repo
stringlengths 2
152
⌀ | file
stringlengths 15
239
| code
stringlengths 0
58.4M
| file_length
int64 0
58.4M
| avg_line_length
float64 0
1.81M
| max_line_length
int64 0
12.7M
| extension_type
stringclasses 364
values |
|---|---|---|---|---|---|---|
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_4x4_896x896_80k_isaid.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py', '../_base_/datasets/isaid.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
model = dict(
decode_head=dict(num_classes=16), auxiliary_head=dict(num_classes=16))
| 248 | 34.571429 | 74 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x1024_40k_cityscapes.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py', '../_base_/datasets/cityscapes.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_40k.py'
]
| 164 | 32 | 76 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x1024_40k_dark.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py', '../_base_/datasets/cityscapes.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_40k.py'
]
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=(1920, 1080),
# img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
flip=False,
transforms=[
dict(type='Resize', keep_ratio=True),
dict(type='RandomFlip'),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img']),
])
]
data = dict(
test=dict(
type='DarkZurichDataset',
data_root='data/dark_zurich/',
img_dir='rgb_anon/val/night/GOPR0356',
ann_dir='gt/val/night/GOPR0356',
pipeline=test_pipeline))
| 967 | 31.266667 | 77 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x1024_40k_night_driving.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py', '../_base_/datasets/cityscapes.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_40k.py'
]
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=(1920, 1080),
# img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
flip=False,
transforms=[
dict(type='Resize', keep_ratio=True),
dict(type='RandomFlip'),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img']),
])
]
data = dict(
test=dict(
type='NightDrivingDataset',
data_root='data/NighttimeDrivingTest/',
img_dir='leftImg8bit/test/night',
ann_dir='gtCoarse_daytime_trainvaltest/test/night',
pipeline=test_pipeline))
| 992 | 32.1 | 77 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x1024_80k_cityscapes.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py', '../_base_/datasets/cityscapes.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
| 164 | 32 | 76 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x1024_80k_dark.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py', '../_base_/datasets/cityscapes.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=(1920, 1080),
# img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
flip=False,
transforms=[
dict(type='Resize', keep_ratio=True),
dict(type='RandomFlip'),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img']),
])
]
data = dict(
test=dict(
type='DarkZurichDataset',
data_root='data/dark_zurich/',
img_dir='rgb_anon/val/night/GOPR0356',
ann_dir='gt/val/night/GOPR0356',
pipeline=test_pipeline))
| 968 | 30.258065 | 77 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x1024_80k_night_driving.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py', '../_base_/datasets/cityscapes.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=(1920, 1080),
# img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
flip=False,
transforms=[
dict(type='Resize', keep_ratio=True),
dict(type='RandomFlip'),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img']),
])
]
data = dict(
test=dict(
type='NightDrivingDataset',
data_root='data/NighttimeDrivingTest/',
img_dir='leftImg8bit/test/night',
ann_dir='gtCoarse_daytime_trainvaltest/test/night',
pipeline=test_pipeline))
| 992 | 32.1 | 77 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x512_160k_ade20k.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py', '../_base_/datasets/ade20k.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_160k.py'
]
model = dict(
decode_head=dict(num_classes=150), auxiliary_head=dict(num_classes=150))
| 252 | 35.142857 | 76 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x512_20k_voc12aug.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py',
'../_base_/datasets/pascal_voc12_aug.py', '../_base_/default_runtime.py',
'../_base_/schedules/schedule_20k.py'
]
model = dict(
decode_head=dict(num_classes=21), auxiliary_head=dict(num_classes=21))
| 263 | 32 | 77 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x512_40k_voc12aug.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py',
'../_base_/datasets/pascal_voc12_aug.py', '../_base_/default_runtime.py',
'../_base_/schedules/schedule_40k.py'
]
model = dict(
decode_head=dict(num_classes=21), auxiliary_head=dict(num_classes=21))
| 263 | 32 | 77 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x512_4x4_160k_coco-stuff164k.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py',
'../_base_/datasets/coco-stuff164k.py', '../_base_/default_runtime.py',
'../_base_/schedules/schedule_160k.py'
]
model = dict(
decode_head=dict(num_classes=171), auxiliary_head=dict(num_classes=171))
| 264 | 32.125 | 76 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x512_4x4_20k_coco-stuff10k.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py', '../_base_/datasets/coco-stuff10k.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_20k.py'
]
model = dict(
decode_head=dict(num_classes=171), auxiliary_head=dict(num_classes=171))
| 258 | 36 | 79 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x512_4x4_320k_coco-stuff164k.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py',
'../_base_/datasets/coco-stuff164k.py', '../_base_/default_runtime.py',
'../_base_/schedules/schedule_320k.py'
]
model = dict(
decode_head=dict(num_classes=171), auxiliary_head=dict(num_classes=171))
| 264 | 32.125 | 76 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x512_4x4_40k_coco-stuff10k.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py', '../_base_/datasets/coco-stuff10k.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_40k.py'
]
model = dict(
decode_head=dict(num_classes=171), auxiliary_head=dict(num_classes=171))
| 258 | 36 | 79 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x512_4x4_80k_coco-stuff164k.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py',
'../_base_/datasets/coco-stuff164k.py', '../_base_/default_runtime.py',
'../_base_/schedules/schedule_80k.py'
]
model = dict(
decode_head=dict(num_classes=171), auxiliary_head=dict(num_classes=171))
| 263 | 32 | 76 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x512_80k_ade20k.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py', '../_base_/datasets/ade20k.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
model = dict(
decode_head=dict(num_classes=150), auxiliary_head=dict(num_classes=150))
| 251 | 35 | 76 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_512x512_80k_loveda.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py', '../_base_/datasets/loveda.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
model = dict(
decode_head=dict(num_classes=7), auxiliary_head=dict(num_classes=7))
| 247 | 34.428571 | 73 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_769x769_40k_cityscapes.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py',
'../_base_/datasets/cityscapes_769x769.py', '../_base_/default_runtime.py',
'../_base_/schedules/schedule_40k.py'
]
model = dict(
decode_head=dict(align_corners=True),
auxiliary_head=dict(align_corners=True),
test_cfg=dict(mode='slide', crop_size=(769, 769), stride=(513, 513)))
| 351 | 34.2 | 79 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_769x769_80k_cityscapes.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py',
'../_base_/datasets/cityscapes_769x769.py', '../_base_/default_runtime.py',
'../_base_/schedules/schedule_80k.py'
]
model = dict(
decode_head=dict(align_corners=True),
auxiliary_head=dict(align_corners=True),
test_cfg=dict(mode='slide', crop_size=(769, 769), stride=(513, 513)))
| 351 | 34.2 | 79 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50-d8_rsb-pretrain_512x1024_adamw_80k_cityscapes.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py', '../_base_/datasets/cityscapes.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
checkpoint = 'https://download.openmmlab.com/mmclassification/v0/resnet/resnet50_8xb256-rsb-a1-600e_in1k_20211228-20e21305.pth' # noqa
model = dict(
pretrained=None,
backbone=dict(
type='ResNet',
init_cfg=dict(
type='Pretrained', prefix='backbone.', checkpoint=checkpoint)))
optimizer = dict(_delete_=True, type='AdamW', lr=0.0005, weight_decay=0.05)
optimizer_config = dict(grad_clip=dict(max_norm=1, norm_type=2))
# learning policy
lr_config = dict(
_delete_=True,
policy='step',
warmup='linear',
warmup_iters=1000,
warmup_ratio=0.001,
step=[60000, 72000],
by_epoch=False)
| 805 | 32.583333 | 135 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50b-d32_512x1024_80k_cityscapes.py | _base_ = [
'../_base_/models/pspnet_r50-d8.py', '../_base_/datasets/cityscapes.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
model = dict(
pretrained='torchvision://resnet50',
backbone=dict(type='ResNet', dilations=(1, 1, 2, 4), strides=(1, 2, 2, 2)))
| 299 | 36.5 | 79 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50b-d8_512x1024_80k_cityscapes.py | _base_ = './pspnet_r50-d8_512x1024_80k_cityscapes.py'
model = dict(pretrained='torchvision://resnet50', backbone=dict(type='ResNet'))
| 134 | 44 | 79 | py |
mmsegmentation | mmsegmentation-master/configs/pspnet/pspnet_r50b-d8_769x769_80k_cityscapes.py | _base_ = './pspnet_r50-d8_769x769_80k_cityscapes.py'
model = dict(pretrained='torchvision://resnet50', backbone=dict(type='ResNet'))
| 133 | 43.666667 | 79 | py |
mmsegmentation | mmsegmentation-master/configs/resnest/README.md | # ResNeSt
[ResNeSt: Split-Attention Networks](https://arxiv.org/abs/2004.08955)
## Introduction
<!-- [BACKBONE] -->
<a href="https://github.com/zhanghang1989/ResNeSt">Official Repo</a>
<a href="https://github.com/open-mmlab/mmsegmentation/blob/v0.17.0/mmseg/models/backbones/resnest.py#L271">Code Snippet</a>
## Abstract
<!-- [ABSTRACT] -->
It is well known that featuremap attention and multi-path representation are important for visual recognition. In this paper, we present a modularized architecture, which applies the channel-wise attention on different network branches to leverage their success in capturing cross-feature interactions and learning diverse representations. Our design results in a simple and unified computation block, which can be parameterized using only a few variables. Our model, named ResNeSt, outperforms EfficientNet in accuracy and latency trade-off on image classification. In addition, ResNeSt has achieved superior transfer learning results on several public benchmarks serving as the backbone, and has been adopted by the winning entries of COCO-LVIS challenge. The source code for complete system and pretrained models are publicly available.
<!-- [IMAGE] -->
<div align=center>
<img src="https://user-images.githubusercontent.com/24582831/142902526-3cf33345-7e40-47a6-985e-4381857e21df.png" width="60%"/>
</div>
## Citation
```bibtex
@article{zhang2020resnest,
title={ResNeSt: Split-Attention Networks},
author={Zhang, Hang and Wu, Chongruo and Zhang, Zhongyue and Zhu, Yi and Zhang, Zhi and Lin, Haibin and Sun, Yue and He, Tong and Muller, Jonas and Manmatha, R. and Li, Mu and Smola, Alexander},
journal={arXiv preprint arXiv:2004.08955},
year={2020}
}
```
## Results and models
### Cityscapes
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| ---------- | -------- | --------- | ------: | -------: | -------------- | ----: | ------------- | ----------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| FCN | S-101-D8 | 512x1024 | 80000 | 11.4 | 2.39 | 77.56 | 78.98 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/resnest/fcn_s101-d8_512x1024_80k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/fcn_s101-d8_512x1024_80k_cityscapes/fcn_s101-d8_512x1024_80k_cityscapes_20200807_140631-f8d155b3.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/fcn_s101-d8_512x1024_80k_cityscapes/fcn_s101-d8_512x1024_80k_cityscapes-20200807_140631.log.json) |
| PSPNet | S-101-D8 | 512x1024 | 80000 | 11.8 | 2.52 | 78.57 | 79.19 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/resnest/pspnet_s101-d8_512x1024_80k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/pspnet_s101-d8_512x1024_80k_cityscapes/pspnet_s101-d8_512x1024_80k_cityscapes_20200807_140631-c75f3b99.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/pspnet_s101-d8_512x1024_80k_cityscapes/pspnet_s101-d8_512x1024_80k_cityscapes-20200807_140631.log.json) |
| DeepLabV3 | S-101-D8 | 512x1024 | 80000 | 11.9 | 1.88 | 79.67 | 80.51 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/resnest/deeplabv3_s101-d8_512x1024_80k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/deeplabv3_s101-d8_512x1024_80k_cityscapes/deeplabv3_s101-d8_512x1024_80k_cityscapes_20200807_144429-b73c4270.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/deeplabv3_s101-d8_512x1024_80k_cityscapes/deeplabv3_s101-d8_512x1024_80k_cityscapes-20200807_144429.log.json) |
| DeepLabV3+ | S-101-D8 | 512x1024 | 80000 | 13.2 | 2.36 | 79.62 | 80.27 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/resnest/deeplabv3plus_s101-d8_512x1024_80k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/deeplabv3plus_s101-d8_512x1024_80k_cityscapes/deeplabv3plus_s101-d8_512x1024_80k_cityscapes_20200807_144429-1239eb43.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/deeplabv3plus_s101-d8_512x1024_80k_cityscapes/deeplabv3plus_s101-d8_512x1024_80k_cityscapes-20200807_144429.log.json) |
### ADE20K
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| ---------- | -------- | --------- | ------: | -------: | -------------- | ----: | ------------- | ------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| FCN | S-101-D8 | 512x512 | 160000 | 14.2 | 12.86 | 45.62 | 46.16 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/resnest/fcn_s101-d8_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/fcn_s101-d8_512x512_160k_ade20k/fcn_s101-d8_512x512_160k_ade20k_20200807_145416-d3160329.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/fcn_s101-d8_512x512_160k_ade20k/fcn_s101-d8_512x512_160k_ade20k-20200807_145416.log.json) |
| PSPNet | S-101-D8 | 512x512 | 160000 | 14.2 | 13.02 | 45.44 | 46.28 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/resnest/pspnet_s101-d8_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/pspnet_s101-d8_512x512_160k_ade20k/pspnet_s101-d8_512x512_160k_ade20k_20200807_145416-a6daa92a.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/pspnet_s101-d8_512x512_160k_ade20k/pspnet_s101-d8_512x512_160k_ade20k-20200807_145416.log.json) |
| DeepLabV3 | S-101-D8 | 512x512 | 160000 | 14.6 | 9.28 | 45.71 | 46.59 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/resnest/deeplabv3_s101-d8_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/deeplabv3_s101-d8_512x512_160k_ade20k/deeplabv3_s101-d8_512x512_160k_ade20k_20200807_144503-17ecabe5.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/deeplabv3_s101-d8_512x512_160k_ade20k/deeplabv3_s101-d8_512x512_160k_ade20k-20200807_144503.log.json) |
| DeepLabV3+ | S-101-D8 | 512x512 | 160000 | 16.2 | 11.96 | 46.47 | 47.27 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/resnest/deeplabv3plus_s101-d8_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/deeplabv3plus_s101-d8_512x512_160k_ade20k/deeplabv3plus_s101-d8_512x512_160k_ade20k_20200807_144503-27b26226.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/resnest/deeplabv3plus_s101-d8_512x512_160k_ade20k/deeplabv3plus_s101-d8_512x512_160k_ade20k-20200807_144503.log.json) |
| 8,976 | 162.218182 | 839 | md |
mmsegmentation | mmsegmentation-master/configs/resnest/deeplabv3_s101-d8_512x1024_80k_cityscapes.py | _base_ = '../deeplabv3/deeplabv3_r101-d8_512x1024_80k_cityscapes.py'
model = dict(
pretrained='open-mmlab://resnest101',
backbone=dict(
type='ResNeSt',
stem_channels=128,
radix=2,
reduction_factor=4,
avg_down_stride=True))
| 271 | 26.2 | 68 | py |
mmsegmentation | mmsegmentation-master/configs/resnest/deeplabv3_s101-d8_512x512_160k_ade20k.py | _base_ = '../deeplabv3/deeplabv3_r101-d8_512x512_160k_ade20k.py'
model = dict(
pretrained='open-mmlab://resnest101',
backbone=dict(
type='ResNeSt',
stem_channels=128,
radix=2,
reduction_factor=4,
avg_down_stride=True))
| 267 | 25.8 | 64 | py |
mmsegmentation | mmsegmentation-master/configs/resnest/deeplabv3plus_s101-d8_512x1024_80k_cityscapes.py | _base_ = '../deeplabv3plus/deeplabv3plus_r101-d8_512x1024_80k_cityscapes.py'
model = dict(
pretrained='open-mmlab://resnest101',
backbone=dict(
type='ResNeSt',
stem_channels=128,
radix=2,
reduction_factor=4,
avg_down_stride=True))
| 279 | 27 | 76 | py |
mmsegmentation | mmsegmentation-master/configs/resnest/deeplabv3plus_s101-d8_512x512_160k_ade20k.py | _base_ = '../deeplabv3plus/deeplabv3plus_r101-d8_512x512_160k_ade20k.py'
model = dict(
pretrained='open-mmlab://resnest101',
backbone=dict(
type='ResNeSt',
stem_channels=128,
radix=2,
reduction_factor=4,
avg_down_stride=True))
| 275 | 26.6 | 72 | py |
mmsegmentation | mmsegmentation-master/configs/resnest/fcn_s101-d8_512x1024_80k_cityscapes.py | _base_ = '../fcn/fcn_r101-d8_512x1024_80k_cityscapes.py'
model = dict(
pretrained='open-mmlab://resnest101',
backbone=dict(
type='ResNeSt',
stem_channels=128,
radix=2,
reduction_factor=4,
avg_down_stride=True))
| 259 | 25 | 56 | py |
mmsegmentation | mmsegmentation-master/configs/resnest/fcn_s101-d8_512x512_160k_ade20k.py | _base_ = '../fcn/fcn_r101-d8_512x512_160k_ade20k.py'
model = dict(
pretrained='open-mmlab://resnest101',
backbone=dict(
type='ResNeSt',
stem_channels=128,
radix=2,
reduction_factor=4,
avg_down_stride=True))
| 255 | 24.6 | 52 | py |
mmsegmentation | mmsegmentation-master/configs/resnest/pspnet_s101-d8_512x1024_80k_cityscapes.py | _base_ = '../pspnet/pspnet_r101-d8_512x1024_80k_cityscapes.py'
model = dict(
pretrained='open-mmlab://resnest101',
backbone=dict(
type='ResNeSt',
stem_channels=128,
radix=2,
reduction_factor=4,
avg_down_stride=True))
| 265 | 25.6 | 62 | py |
mmsegmentation | mmsegmentation-master/configs/resnest/pspnet_s101-d8_512x512_160k_ade20k.py | _base_ = '../pspnet/pspnet_r101-d8_512x512_160k_ade20k.py'
model = dict(
pretrained='open-mmlab://resnest101',
backbone=dict(
type='ResNeSt',
stem_channels=128,
radix=2,
reduction_factor=4,
avg_down_stride=True))
| 261 | 25.2 | 58 | py |
mmsegmentation | mmsegmentation-master/configs/resnest/resnest.yml | Models:
- Name: fcn_s101-d8_512x1024_80k_cityscapes
In Collection: FCN
Metadata:
backbone: S-101-D8
crop size: (512,1024)
lr schd: 80000
inference time (ms/im):
- value: 418.41
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,1024)
Training Memory (GB): 11.4
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 77.56
mIoU(ms+flip): 78.98
Config: configs/resnest/fcn_s101-d8_512x1024_80k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/resnest/fcn_s101-d8_512x1024_80k_cityscapes/fcn_s101-d8_512x1024_80k_cityscapes_20200807_140631-f8d155b3.pth
- Name: pspnet_s101-d8_512x1024_80k_cityscapes
In Collection: PSPNet
Metadata:
backbone: S-101-D8
crop size: (512,1024)
lr schd: 80000
inference time (ms/im):
- value: 396.83
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,1024)
Training Memory (GB): 11.8
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 78.57
mIoU(ms+flip): 79.19
Config: configs/resnest/pspnet_s101-d8_512x1024_80k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/resnest/pspnet_s101-d8_512x1024_80k_cityscapes/pspnet_s101-d8_512x1024_80k_cityscapes_20200807_140631-c75f3b99.pth
- Name: deeplabv3_s101-d8_512x1024_80k_cityscapes
In Collection: DeepLabV3
Metadata:
backbone: S-101-D8
crop size: (512,1024)
lr schd: 80000
inference time (ms/im):
- value: 531.91
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,1024)
Training Memory (GB): 11.9
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 79.67
mIoU(ms+flip): 80.51
Config: configs/resnest/deeplabv3_s101-d8_512x1024_80k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/resnest/deeplabv3_s101-d8_512x1024_80k_cityscapes/deeplabv3_s101-d8_512x1024_80k_cityscapes_20200807_144429-b73c4270.pth
- Name: deeplabv3plus_s101-d8_512x1024_80k_cityscapes
In Collection: DeepLabV3+
Metadata:
backbone: S-101-D8
crop size: (512,1024)
lr schd: 80000
inference time (ms/im):
- value: 423.73
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,1024)
Training Memory (GB): 13.2
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 79.62
mIoU(ms+flip): 80.27
Config: configs/resnest/deeplabv3plus_s101-d8_512x1024_80k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/resnest/deeplabv3plus_s101-d8_512x1024_80k_cityscapes/deeplabv3plus_s101-d8_512x1024_80k_cityscapes_20200807_144429-1239eb43.pth
- Name: fcn_s101-d8_512x512_160k_ade20k
In Collection: FCN
Metadata:
backbone: S-101-D8
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 77.76
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 14.2
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 45.62
mIoU(ms+flip): 46.16
Config: configs/resnest/fcn_s101-d8_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/resnest/fcn_s101-d8_512x512_160k_ade20k/fcn_s101-d8_512x512_160k_ade20k_20200807_145416-d3160329.pth
- Name: pspnet_s101-d8_512x512_160k_ade20k
In Collection: PSPNet
Metadata:
backbone: S-101-D8
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 76.8
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 14.2
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 45.44
mIoU(ms+flip): 46.28
Config: configs/resnest/pspnet_s101-d8_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/resnest/pspnet_s101-d8_512x512_160k_ade20k/pspnet_s101-d8_512x512_160k_ade20k_20200807_145416-a6daa92a.pth
- Name: deeplabv3_s101-d8_512x512_160k_ade20k
In Collection: DeepLabV3
Metadata:
backbone: S-101-D8
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 107.76
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 14.6
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 45.71
mIoU(ms+flip): 46.59
Config: configs/resnest/deeplabv3_s101-d8_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/resnest/deeplabv3_s101-d8_512x512_160k_ade20k/deeplabv3_s101-d8_512x512_160k_ade20k_20200807_144503-17ecabe5.pth
- Name: deeplabv3plus_s101-d8_512x512_160k_ade20k
In Collection: DeepLabV3+
Metadata:
backbone: S-101-D8
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 83.61
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 16.2
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 46.47
mIoU(ms+flip): 47.27
Config: configs/resnest/deeplabv3plus_s101-d8_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/resnest/deeplabv3plus_s101-d8_512x512_160k_ade20k/deeplabv3plus_s101-d8_512x512_160k_ade20k_20200807_144503-27b26226.pth
| 5,664 | 30.825843 | 190 | yml |
mmsegmentation | mmsegmentation-master/configs/segformer/README.md | # SegFormer
[SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers](https://arxiv.org/abs/2105.15203)
## Introduction
<!-- [ALGORITHM] -->
<a href="https://github.com/NVlabs/SegFormer">Official Repo</a>
<a href="https://github.com/open-mmlab/mmsegmentation/blob/v0.17.0/mmseg/models/backbones/mit.py#L246">Code Snippet</a>
## Abstract
<!-- [ABSTRACT] -->
We present SegFormer, a simple, efficient yet powerful semantic segmentation framework which unifies Transformers with lightweight multilayer perception (MLP) decoders. SegFormer has two appealing features: 1) SegFormer comprises a novel hierarchically structured Transformer encoder which outputs multiscale features. It does not need positional encoding, thereby avoiding the interpolation of positional codes which leads to decreased performance when the testing resolution differs from training. 2) SegFormer avoids complex decoders. The proposed MLP decoder aggregates information from different layers, and thus combining both local attention and global attention to render powerful representations. We show that this simple and lightweight design is the key to efficient segmentation on Transformers. We scale our approach up to obtain a series of models from SegFormer-B0 to SegFormer-B5, reaching significantly better performance and efficiency than previous counterparts. For example, SegFormer-B4 achieves 50.3% mIoU on ADE20K with 64M parameters, being 5x smaller and 2.2% better than the previous best method. Our best model, SegFormer-B5, achieves 84.0% mIoU on Cityscapes validation set and shows excellent zero-shot robustness on Cityscapes-C. Code will be released at: [this http URL](https://github.com/NVlabs/SegFormer).
<!-- [IMAGE] -->
<div align=center>
<img src="https://user-images.githubusercontent.com/24582831/142902600-e188073e-5744-4ba9-8dbf-9316e55c74aa.png" width="70%"/>
</div>
## Citation
```bibtex
@article{xie2021segformer,
title={SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers},
author={Xie, Enze and Wang, Wenhai and Yu, Zhiding and Anandkumar, Anima and Alvarez, Jose M and Luo, Ping},
journal={arXiv preprint arXiv:2105.15203},
year={2021}
}
```
## Usage
We have provided pretrained models converted from [SegFormer](https://github.com/NVlabs/SegFormer).
If you want to convert keys on your own, we also provide a script [`mit2mmseg.py`](../../tools/model_converters/mit2mmseg.py) in the tools directory to convert the key of models from [the official repo](https://github.com/NVlabs/SegFormer) to MMSegmentation style.
```shell
python tools/model_converters/mit2mmseg.py ${PRETRAIN_PATH} ${STORE_PATH}
```
This script convert model from `PRETRAIN_PATH` and store the converted model in `STORE_PATH`.
## Results and models
### ADE20K
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| --------- | -------- | --------- | ------: | -------: | -------------- | ----: | ------------- | ---------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| Segformer | MIT-B0 | 512x512 | 160000 | 2.1 | 38.17 | 37.85 | 38.97 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segformer/segformer_mit-b0_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b0_512x512_160k_ade20k/segformer_mit-b0_512x512_160k_ade20k_20220617_162207-c00b9603.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b0_512x512_160k_ade20k/segformer_mit-b0_512x512_160k_ade20k_20220617_162207.log.json) |
| Segformer | MIT-B1 | 512x512 | 160000 | 2.6 | 37.80 | 42.13 | 43.74 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segformer/segformer_mit-b1_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b1_512x512_160k_ade20k/segformer_mit-b1_512x512_160k_ade20k_20220620_112037-c3f39e00.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b1_512x512_160k_ade20k/segformer_mit-b1_512x512_160k_ade20k_20220620_112037.log.json) |
| Segformer | MIT-B2 | 512x512 | 160000 | 3.6 | 26.80 | 46.80 | 48.12 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segformer/segformer_mit-b2_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b2_512x512_160k_ade20k/segformer_mit-b2_512x512_160k_ade20k_20220620_114047-64e4feca.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b2_512x512_160k_ade20k/segformer_mit-b2_512x512_160k_ade20k_20220620_114047.log.json) |
| Segformer | MIT-B3 | 512x512 | 160000 | 4.8 | 19.19 | 48.25 | 49.58 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segformer/segformer_mit-b3_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b3_512x512_160k_ade20k/segformer_mit-b3_512x512_160k_ade20k_20220617_162254-3a4b7363.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b3_512x512_160k_ade20k/segformer_mit-b3_512x512_160k_ade20k_20220617_162254.log.json) |
| Segformer | MIT-B4 | 512x512 | 160000 | 6.1 | 14.54 | 49.09 | 50.72 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segformer/segformer_mit-b4_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b4_512x512_160k_ade20k/segformer_mit-b4_512x512_160k_ade20k_20220620_112216-4fa4f58f.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b4_512x512_160k_ade20k/segformer_mit-b4_512x512_160k_ade20k_20220620_112216.log.json) |
| Segformer | MIT-B5 | 512x512 | 160000 | 7.2 | 11.89 | 49.13 | 50.22 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segformer/segformer_mit-b5_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b5_512x512_160k_ade20k/segformer_mit-b5_512x512_160k_ade20k_20210726_145235-94cedf59.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b5_512x512_160k_ade20k/segformer_mit-b5_512x512_160k_ade20k_20210726_145235.log.json) |
| Segformer | MIT-B5 | 640x640 | 160000 | 11.5 | 10.60 | 50.19 | 51.41 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segformer/segformer_mit-b5_640x640_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b5_640x640_160k_ade20k/segformer_mit-b5_640x640_160k_ade20k_20220617_203542-940a6bd8.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b5_640x640_160k_ade20k/segformer_mit-b5_640x640_160k_ade20k_20220617_203542.log.json) |
Evaluation with `AlignedResize`:
| Method | Backbone | Crop Size | Lr schd | mIoU | mIoU(ms+flip) |
| --------- | -------- | --------- | ------: | ----: | ------------- |
| Segformer | MIT-B0 | 512x512 | 160000 | 38.55 | 39.03 |
| Segformer | MIT-B1 | 512x512 | 160000 | 43.26 | 44.11 |
| Segformer | MIT-B2 | 512x512 | 160000 | 47.46 | 48.16 |
| Segformer | MIT-B3 | 512x512 | 160000 | 49.27 | 49.94 |
| Segformer | MIT-B4 | 512x512 | 160000 | 50.23 | 51.10 |
| Segformer | MIT-B5 | 512x512 | 160000 | 50.08 | 50.72 |
| Segformer | MIT-B5 | 640x640 | 160000 | 51.13 | 51.66 |
### Cityscapes
The lower fps result is caused by the sliding window inference scheme (window size:1024x1024).
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| --------- | -------- | --------- | ------: | -------: | -------------- | ----: | ------------- | -------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Segformer | MIT-B0 | 1024x1024 | 160000 | 3.64 | 4.74 | 76.54 | 78.22 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segformer/segformer_mit-b0_8x1_1024x1024_160k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b0_8x1_1024x1024_160k_cityscapes/segformer_mit-b0_8x1_1024x1024_160k_cityscapes_20211208_101857-e7f88502.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b0_8x1_1024x1024_160k_cityscapes/segformer_mit-b0_8x1_1024x1024_160k_cityscapes_20211208_101857.log.json) |
| Segformer | MIT-B1 | 1024x1024 | 160000 | 4.49 | 4.3 | 78.56 | 79.73 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segformer/segformer_mit-b1_8x1_1024x1024_160k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b1_8x1_1024x1024_160k_cityscapes/segformer_mit-b1_8x1_1024x1024_160k_cityscapes_20211208_064213-655c7b3f.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b1_8x1_1024x1024_160k_cityscapes/segformer_mit-b1_8x1_1024x1024_160k_cityscapes_20211208_064213.log.json) |
| Segformer | MIT-B2 | 1024x1024 | 160000 | 7.42 | 3.36 | 81.08 | 82.18 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segformer/segformer_mit-b2_8x1_1024x1024_160k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b2_8x1_1024x1024_160k_cityscapes/segformer_mit-b2_8x1_1024x1024_160k_cityscapes_20211207_134205-6096669a.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b2_8x1_1024x1024_160k_cityscapes/segformer_mit-b2_8x1_1024x1024_160k_cityscapes_20211207_134205.log.json) |
| Segformer | MIT-B3 | 1024x1024 | 160000 | 10.86 | 2.53 | 81.94 | 83.14 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segformer/segformer_mit-b3_8x1_1024x1024_160k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b3_8x1_1024x1024_160k_cityscapes/segformer_mit-b3_8x1_1024x1024_160k_cityscapes_20211206_224823-a8f8a177.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b3_8x1_1024x1024_160k_cityscapes/segformer_mit-b3_8x1_1024x1024_160k_cityscapes_20211206_224823.log.json) |
| Segformer | MIT-B4 | 1024x1024 | 160000 | 15.07 | 1.88 | 81.89 | 83.38 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segformer/segformer_mit-b4_8x1_1024x1024_160k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b4_8x1_1024x1024_160k_cityscapes/segformer_mit-b4_8x1_1024x1024_160k_cityscapes_20211207_080709-07f6c333.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b4_8x1_1024x1024_160k_cityscapes/segformer_mit-b4_8x1_1024x1024_160k_cityscapes_20211207_080709.log.json) |
| Segformer | MIT-B5 | 1024x1024 | 160000 | 18.00 | 1.39 | 82.25 | 83.48 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segformer/segformer_mit-b5_8x1_1024x1024_160k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b5_8x1_1024x1024_160k_cityscapes/segformer_mit-b5_8x1_1024x1024_160k_cityscapes_20211206_072934-87a052ec.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b5_8x1_1024x1024_160k_cityscapes/segformer_mit-b5_8x1_1024x1024_160k_cityscapes_20211206_072934.log.json) |
Note:
Original SegFormer paper uses different `test_pipeline` and image ratios in `ms+flip`. If you want to cite SegFormer original results as benchmark you may modify settings as below:
- We replace `AlignedResize` in original implementation to `Resize + ResizeToMultiple`. If you want to test by
using `AlignedResize`, you can change the dataset pipeline like this:
```python
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=(2048, 512),
# img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
flip=False,
transforms=[
dict(type='Resize', keep_ratio=True),
# resize image to multiple of 32, improve SegFormer by 0.5-1.0 mIoU.
dict(type='ResizeToMultiple', size_divisor=32),
dict(type='RandomFlip'),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img']),
])
]
```
- Different from default setting of `ms+flip`, SegFormer original repo adopts [different image ratios](https://github.com/NVlabs/SegFormer/blob/master/tools/test.py#L97-L101) for ADE20K dataset. To re-implement numerical results of `ms+flip`, you can change image ratios in `tools/test.py` like this:
```python
if args.aug_test:
if cfg.data.test.type == 'ADE20KDataset':
# hard code index
cfg.data.test.pipeline[1].img_ratios = [
0.75, 0.875, 1.0, 1.125, 1.25
]
```
- Training of SegFormer is not very stable, which is sensitive to random seeds.
- We use default training setting in MMSegmentation rather than `RepeatDataset` adopted in SegFormer official repo to accelerate [training](https://github.com/NVlabs/SegFormer/blob/master/local_configs/_base_/datasets/ade20k_repeat.py#L38-L39), here is its related [issue](https://github.com/NVlabs/SegFormer/issues/25).
| 15,635 | 121.15625 | 1,339 | md |
mmsegmentation | mmsegmentation-master/configs/segformer/segformer.yml | Collections:
- Name: Segformer
Metadata:
Training Data:
- ADE20K
- Cityscapes
Paper:
URL: https://arxiv.org/abs/2105.15203
Title: 'SegFormer: Simple and Efficient Design for Semantic Segmentation with
Transformers'
README: configs/segformer/README.md
Code:
URL: https://github.com/open-mmlab/mmsegmentation/blob/v0.17.0/mmseg/models/backbones/mit.py#L246
Version: v0.17.0
Converted From:
Code: https://github.com/NVlabs/SegFormer
Models:
- Name: segformer_mit-b0_512x512_160k_ade20k
In Collection: Segformer
Metadata:
backbone: MIT-B0
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 26.2
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 2.1
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 37.85
mIoU(ms+flip): 38.97
Config: configs/segformer/segformer_mit-b0_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b0_512x512_160k_ade20k/segformer_mit-b0_512x512_160k_ade20k_20220617_162207-c00b9603.pth
- Name: segformer_mit-b1_512x512_160k_ade20k
In Collection: Segformer
Metadata:
backbone: MIT-B1
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 26.46
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 2.6
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 42.13
mIoU(ms+flip): 43.74
Config: configs/segformer/segformer_mit-b1_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b1_512x512_160k_ade20k/segformer_mit-b1_512x512_160k_ade20k_20220620_112037-c3f39e00.pth
- Name: segformer_mit-b2_512x512_160k_ade20k
In Collection: Segformer
Metadata:
backbone: MIT-B2
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 37.31
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 3.6
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 46.8
mIoU(ms+flip): 48.12
Config: configs/segformer/segformer_mit-b2_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b2_512x512_160k_ade20k/segformer_mit-b2_512x512_160k_ade20k_20220620_114047-64e4feca.pth
- Name: segformer_mit-b3_512x512_160k_ade20k
In Collection: Segformer
Metadata:
backbone: MIT-B3
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 52.11
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 4.8
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 48.25
mIoU(ms+flip): 49.58
Config: configs/segformer/segformer_mit-b3_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b3_512x512_160k_ade20k/segformer_mit-b3_512x512_160k_ade20k_20220617_162254-3a4b7363.pth
- Name: segformer_mit-b4_512x512_160k_ade20k
In Collection: Segformer
Metadata:
backbone: MIT-B4
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 68.78
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 6.1
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 49.09
mIoU(ms+flip): 50.72
Config: configs/segformer/segformer_mit-b4_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b4_512x512_160k_ade20k/segformer_mit-b4_512x512_160k_ade20k_20220620_112216-4fa4f58f.pth
- Name: segformer_mit-b5_512x512_160k_ade20k
In Collection: Segformer
Metadata:
backbone: MIT-B5
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 84.1
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 7.2
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 49.13
mIoU(ms+flip): 50.22
Config: configs/segformer/segformer_mit-b5_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b5_512x512_160k_ade20k/segformer_mit-b5_512x512_160k_ade20k_20210726_145235-94cedf59.pth
- Name: segformer_mit-b5_640x640_160k_ade20k
In Collection: Segformer
Metadata:
backbone: MIT-B5
crop size: (640,640)
lr schd: 160000
inference time (ms/im):
- value: 94.34
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (640,640)
Training Memory (GB): 11.5
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 50.19
mIoU(ms+flip): 51.41
Config: configs/segformer/segformer_mit-b5_640x640_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b5_640x640_160k_ade20k/segformer_mit-b5_640x640_160k_ade20k_20220617_203542-940a6bd8.pth
- Name: segformer_mit-b0_8x1_1024x1024_160k_cityscapes
In Collection: Segformer
Metadata:
backbone: MIT-B0
crop size: (1024,1024)
lr schd: 160000
inference time (ms/im):
- value: 210.97
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (1024,1024)
Training Memory (GB): 3.64
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 76.54
mIoU(ms+flip): 78.22
Config: configs/segformer/segformer_mit-b0_8x1_1024x1024_160k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b0_8x1_1024x1024_160k_cityscapes/segformer_mit-b0_8x1_1024x1024_160k_cityscapes_20211208_101857-e7f88502.pth
- Name: segformer_mit-b1_8x1_1024x1024_160k_cityscapes
In Collection: Segformer
Metadata:
backbone: MIT-B1
crop size: (1024,1024)
lr schd: 160000
inference time (ms/im):
- value: 232.56
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (1024,1024)
Training Memory (GB): 4.49
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 78.56
mIoU(ms+flip): 79.73
Config: configs/segformer/segformer_mit-b1_8x1_1024x1024_160k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b1_8x1_1024x1024_160k_cityscapes/segformer_mit-b1_8x1_1024x1024_160k_cityscapes_20211208_064213-655c7b3f.pth
- Name: segformer_mit-b2_8x1_1024x1024_160k_cityscapes
In Collection: Segformer
Metadata:
backbone: MIT-B2
crop size: (1024,1024)
lr schd: 160000
inference time (ms/im):
- value: 297.62
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (1024,1024)
Training Memory (GB): 7.42
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 81.08
mIoU(ms+flip): 82.18
Config: configs/segformer/segformer_mit-b2_8x1_1024x1024_160k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b2_8x1_1024x1024_160k_cityscapes/segformer_mit-b2_8x1_1024x1024_160k_cityscapes_20211207_134205-6096669a.pth
- Name: segformer_mit-b3_8x1_1024x1024_160k_cityscapes
In Collection: Segformer
Metadata:
backbone: MIT-B3
crop size: (1024,1024)
lr schd: 160000
inference time (ms/im):
- value: 395.26
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (1024,1024)
Training Memory (GB): 10.86
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 81.94
mIoU(ms+flip): 83.14
Config: configs/segformer/segformer_mit-b3_8x1_1024x1024_160k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b3_8x1_1024x1024_160k_cityscapes/segformer_mit-b3_8x1_1024x1024_160k_cityscapes_20211206_224823-a8f8a177.pth
- Name: segformer_mit-b4_8x1_1024x1024_160k_cityscapes
In Collection: Segformer
Metadata:
backbone: MIT-B4
crop size: (1024,1024)
lr schd: 160000
inference time (ms/im):
- value: 531.91
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (1024,1024)
Training Memory (GB): 15.07
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 81.89
mIoU(ms+flip): 83.38
Config: configs/segformer/segformer_mit-b4_8x1_1024x1024_160k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b4_8x1_1024x1024_160k_cityscapes/segformer_mit-b4_8x1_1024x1024_160k_cityscapes_20211207_080709-07f6c333.pth
- Name: segformer_mit-b5_8x1_1024x1024_160k_cityscapes
In Collection: Segformer
Metadata:
backbone: MIT-B5
crop size: (1024,1024)
lr schd: 160000
inference time (ms/im):
- value: 719.42
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (1024,1024)
Training Memory (GB): 18.0
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 82.25
mIoU(ms+flip): 83.48
Config: configs/segformer/segformer_mit-b5_8x1_1024x1024_160k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segformer/segformer_mit-b5_8x1_1024x1024_160k_cityscapes/segformer_mit-b5_8x1_1024x1024_160k_cityscapes_20211206_072934-87a052ec.pth
| 9,886 | 31.523026 | 194 | yml |
mmsegmentation | mmsegmentation-master/configs/segformer/segformer_mit-b0_512x512_160k_ade20k.py | _base_ = [
'../_base_/models/segformer_mit-b0.py', '../_base_/datasets/ade20k.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_160k.py'
]
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b0_20220624-7e0fe6dd.pth' # noqa
model = dict(pretrained=checkpoint, decode_head=dict(num_classes=150))
# optimizer
optimizer = dict(
_delete_=True,
type='AdamW',
lr=0.00006,
betas=(0.9, 0.999),
weight_decay=0.01,
paramwise_cfg=dict(
custom_keys={
'pos_block': dict(decay_mult=0.),
'norm': dict(decay_mult=0.),
'head': dict(lr_mult=10.)
}))
lr_config = dict(
_delete_=True,
policy='poly',
warmup='linear',
warmup_iters=1500,
warmup_ratio=1e-6,
power=1.0,
min_lr=0.0,
by_epoch=False)
data = dict(samples_per_gpu=2, workers_per_gpu=2)
| 899 | 24.714286 | 121 | py |
mmsegmentation | mmsegmentation-master/configs/segformer/segformer_mit-b0_8x1_1024x1024_160k_cityscapes.py | _base_ = [
'../_base_/models/segformer_mit-b0.py',
'../_base_/datasets/cityscapes_1024x1024.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_160k.py'
]
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b0_20220624-7e0fe6dd.pth' # noqa
model = dict(
backbone=dict(init_cfg=dict(type='Pretrained', checkpoint=checkpoint)),
test_cfg=dict(mode='slide', crop_size=(1024, 1024), stride=(768, 768)))
# optimizer
optimizer = dict(
_delete_=True,
type='AdamW',
lr=0.00006,
betas=(0.9, 0.999),
weight_decay=0.01,
paramwise_cfg=dict(
custom_keys={
'pos_block': dict(decay_mult=0.),
'norm': dict(decay_mult=0.),
'head': dict(lr_mult=10.)
}))
lr_config = dict(
_delete_=True,
policy='poly',
warmup='linear',
warmup_iters=1500,
warmup_ratio=1e-6,
power=1.0,
min_lr=0.0,
by_epoch=False)
data = dict(samples_per_gpu=1, workers_per_gpu=1)
| 1,012 | 25.657895 | 121 | py |
mmsegmentation | mmsegmentation-master/configs/segformer/segformer_mit-b1_512x512_160k_ade20k.py | _base_ = ['./segformer_mit-b0_512x512_160k_ade20k.py']
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b1_20220624-02e5a6a1.pth' # noqa
# model settings
model = dict(
pretrained=checkpoint,
backbone=dict(
embed_dims=64, num_heads=[1, 2, 5, 8], num_layers=[2, 2, 2, 2]),
decode_head=dict(in_channels=[64, 128, 320, 512]))
| 384 | 34 | 121 | py |
mmsegmentation | mmsegmentation-master/configs/segformer/segformer_mit-b1_8x1_1024x1024_160k_cityscapes.py | _base_ = ['./segformer_mit-b0_8x1_1024x1024_160k_cityscapes.py']
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b1_20220624-02e5a6a1.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
embed_dims=64),
decode_head=dict(in_channels=[64, 128, 320, 512]))
| 365 | 39.666667 | 121 | py |
mmsegmentation | mmsegmentation-master/configs/segformer/segformer_mit-b2_512x512_160k_ade20k.py | _base_ = ['./segformer_mit-b0_512x512_160k_ade20k.py']
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b2_20220624-66e8bf70.pth' # noqa
# model settings
model = dict(
pretrained=checkpoint,
backbone=dict(
embed_dims=64, num_heads=[1, 2, 5, 8], num_layers=[3, 4, 6, 3]),
decode_head=dict(in_channels=[64, 128, 320, 512]))
| 384 | 34 | 121 | py |
mmsegmentation | mmsegmentation-master/configs/segformer/segformer_mit-b2_8x1_1024x1024_160k_cityscapes.py | _base_ = ['./segformer_mit-b0_8x1_1024x1024_160k_cityscapes.py']
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b2_20220624-66e8bf70.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
embed_dims=64,
num_layers=[3, 4, 6, 3]),
decode_head=dict(in_channels=[64, 128, 320, 512]))
| 398 | 38.9 | 121 | py |
mmsegmentation | mmsegmentation-master/configs/segformer/segformer_mit-b3_512x512_160k_ade20k.py | _base_ = ['./segformer_mit-b0_512x512_160k_ade20k.py']
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b3_20220624-13b1141c.pth' # noqa
# model settings
model = dict(
pretrained=checkpoint,
backbone=dict(
embed_dims=64, num_heads=[1, 2, 5, 8], num_layers=[3, 4, 18, 3]),
decode_head=dict(in_channels=[64, 128, 320, 512]))
| 385 | 34.090909 | 121 | py |
mmsegmentation | mmsegmentation-master/configs/segformer/segformer_mit-b3_8x1_1024x1024_160k_cityscapes.py | _base_ = ['./segformer_mit-b0_8x1_1024x1024_160k_cityscapes.py']
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b3_20220624-13b1141c.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
embed_dims=64,
num_layers=[3, 4, 18, 3]),
decode_head=dict(in_channels=[64, 128, 320, 512]))
| 399 | 39 | 121 | py |
mmsegmentation | mmsegmentation-master/configs/segformer/segformer_mit-b4_512x512_160k_ade20k.py | _base_ = ['./segformer_mit-b0_512x512_160k_ade20k.py']
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b4_20220624-d588d980.pth' # noqa
# model settings
model = dict(
pretrained=checkpoint,
backbone=dict(
embed_dims=64, num_heads=[1, 2, 5, 8], num_layers=[3, 8, 27, 3]),
decode_head=dict(in_channels=[64, 128, 320, 512]))
| 385 | 34.090909 | 121 | py |
mmsegmentation | mmsegmentation-master/configs/segformer/segformer_mit-b4_8x1_1024x1024_160k_cityscapes.py | _base_ = ['./segformer_mit-b0_8x1_1024x1024_160k_cityscapes.py']
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b4_20220624-d588d980.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
embed_dims=64,
num_layers=[3, 8, 27, 3]),
decode_head=dict(in_channels=[64, 128, 320, 512]))
| 399 | 39 | 121 | py |
mmsegmentation | mmsegmentation-master/configs/segformer/segformer_mit-b5_512x512_160k_ade20k.py | _base_ = ['./segformer_mit-b0_512x512_160k_ade20k.py']
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b5_20220624-658746d9.pth' # noqa
# model settings
model = dict(
pretrained=checkpoint,
backbone=dict(
embed_dims=64, num_heads=[1, 2, 5, 8], num_layers=[3, 6, 40, 3]),
decode_head=dict(in_channels=[64, 128, 320, 512]))
| 385 | 34.090909 | 121 | py |
mmsegmentation | mmsegmentation-master/configs/segformer/segformer_mit-b5_640x640_160k_ade20k.py | _base_ = ['./segformer_mit-b0_512x512_160k_ade20k.py']
# dataset settings
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
crop_size = (640, 640)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadAnnotations', reduce_zero_label=True),
dict(type='Resize', img_scale=(2048, 640), ratio_range=(0.5, 2.0)),
dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
dict(type='RandomFlip', prob=0.5),
dict(type='PhotoMetricDistortion'),
dict(type='Normalize', **img_norm_cfg),
dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img', 'gt_semantic_seg']),
]
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=(2048, 640),
# img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
flip=False,
transforms=[
dict(type='Resize', keep_ratio=True),
dict(type='RandomFlip'),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img']),
])
]
data = dict(
train=dict(pipeline=train_pipeline),
val=dict(pipeline=test_pipeline),
test=dict(pipeline=test_pipeline))
# model settings
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b5_20220624-658746d9.pth' # noqa
model = dict(
pretrained=checkpoint,
backbone=dict(
embed_dims=64, num_heads=[1, 2, 5, 8], num_layers=[3, 6, 40, 3]),
decode_head=dict(in_channels=[64, 128, 320, 512]))
| 1,680 | 35.543478 | 121 | py |
mmsegmentation | mmsegmentation-master/configs/segformer/segformer_mit-b5_8x1_1024x1024_160k_cityscapes.py | _base_ = ['./segformer_mit-b0_8x1_1024x1024_160k_cityscapes.py']
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segformer/mit_b5_20220624-658746d9.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
embed_dims=64,
num_layers=[3, 6, 40, 3]),
decode_head=dict(in_channels=[64, 128, 320, 512]))
| 399 | 39 | 121 | py |
mmsegmentation | mmsegmentation-master/configs/segmenter/README.md | # Segmenter
[Segmenter: Transformer for Semantic Segmentation](https://arxiv.org/abs/2105.05633)
## Introduction
<!-- [ALGORITHM] -->
<a href="https://github.com/rstrudel/segmenter">Official Repo</a>
<a href="https://github.com/open-mmlab/mmsegmentation/blob/v0.21.0/mmseg/models/decode_heads/segmenter_mask_head.py#L15">Code Snippet</a>
## Abstract
<!-- [ABSTRACT] -->
Image segmentation is often ambiguous at the level of individual image patches and requires contextual information to reach label consensus. In this paper we introduce Segmenter, a transformer model for semantic segmentation. In contrast to convolution-based methods, our approach allows to model global context already at the first layer and throughout the network. We build on the recent Vision Transformer (ViT) and extend it to semantic segmentation. To do so, we rely on the output embeddings corresponding to image patches and obtain class labels from these embeddings with a point-wise linear decoder or a mask transformer decoder. We leverage models pre-trained for image classification and show that we can fine-tune them on moderate sized datasets available for semantic segmentation. The linear decoder allows to obtain excellent results already, but the performance can be further improved by a mask transformer generating class masks. We conduct an extensive ablation study to show the impact of the different parameters, in particular the performance is better for large models and small patch sizes. Segmenter attains excellent results for semantic segmentation. It outperforms the state of the art on both ADE20K and Pascal Context datasets and is competitive on Cityscapes.
<!-- [IMAGE] -->
<div align=center>
<img src="https://user-images.githubusercontent.com/24582831/148507554-87eb80bd-02c7-4c31-b102-c6141e231ec8.png" width="70%"/>
</div>
```bibtex
@inproceedings{strudel2021segmenter,
title={Segmenter: Transformer for semantic segmentation},
author={Strudel, Robin and Garcia, Ricardo and Laptev, Ivan and Schmid, Cordelia},
booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
pages={7262--7272},
year={2021}
}
```
## Usage
We have provided pretrained models converted from [ViT-AugReg](https://github.com/rwightman/pytorch-image-models/blob/f55c22bebf9d8afc449d317a723231ef72e0d662/timm/models/vision_transformer.py#L54-L106).
If you want to convert keys on your own to use the pre-trained ViT model from [Segmenter](https://github.com/rstrudel/segmenter), we also provide a script [`vitjax2mmseg.py`](../../tools/model_converters/vitjax2mmseg.py) in the tools directory to convert the key of models from [ViT-AugReg](https://github.com/rwightman/pytorch-image-models/blob/f55c22bebf9d8afc449d317a723231ef72e0d662/timm/models/vision_transformer.py#L54-L106) to MMSegmentation style.
```shell
python tools/model_converters/vitjax2mmseg.py ${PRETRAIN_PATH} ${STORE_PATH}
```
E.g.
```shell
python tools/model_converters/vitjax2mmseg.py \
Ti_16-i21k-300ep-lr_0.001-aug_none-wd_0.03-do_0.0-sd_0.0--imagenet2012-steps_20k-lr_0.03-res_384.npz \
pretrain/vit_tiny_p16_384.pth
```
This script convert model from `PRETRAIN_PATH` and store the converted model in `STORE_PATH`.
In our default setting, pretrained models and their corresponding [ViT-AugReg](https://github.com/rwightman/pytorch-image-models/blob/f55c22bebf9d8afc449d317a723231ef72e0d662/timm/models/vision_transformer.py#L54-L106) models could be defined below:
| pretrained models | original models |
| --------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| vit_tiny_p16_384.pth | ['vit_tiny_patch16_384'](https://storage.googleapis.com/vit_models/augreg/Ti_16-i21k-300ep-lr_0.001-aug_none-wd_0.03-do_0.0-sd_0.0--imagenet2012-steps_20k-lr_0.03-res_384.npz) |
| vit_small_p16_384.pth | ['vit_small_patch16_384'](https://storage.googleapis.com/vit_models/augreg/S_16-i21k-300ep-lr_0.001-aug_light1-wd_0.03-do_0.0-sd_0.0--imagenet2012-steps_20k-lr_0.03-res_384.npz) |
| vit_base_p16_384.pth | ['vit_base_patch16_384'](https://storage.googleapis.com/vit_models/augreg/B_16-i21k-300ep-lr_0.001-aug_medium1-wd_0.1-do_0.0-sd_0.0--imagenet2012-steps_20k-lr_0.01-res_384.npz) |
| vit_large_p16_384.pth | ['vit_large_patch16_384'](https://storage.googleapis.com/vit_models/augreg/L_16-i21k-300ep-lr_0.001-aug_medium1-wd_0.1-do_0.1-sd_0.1--imagenet2012-steps_20k-lr_0.01-res_384.npz) |
## Results and models
### ADE20K
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| ---------------- | -------- | --------- | ------- | -------- | -------------- | ----- | ------------- | -------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Segmenter Mask | ViT-T_16 | 512x512 | 160000 | 1.21 | 27.98 | 39.99 | 40.83 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segmenter/segmenter_vit-t_mask_8x1_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segmenter/segmenter_vit-t_mask_8x1_512x512_160k_ade20k/segmenter_vit-t_mask_8x1_512x512_160k_ade20k_20220105_151706-ffcf7509.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segmenter/segmenter_vit-t_mask_8x1_512x512_160k_ade20k/segmenter_vit-t_mask_8x1_512x512_160k_ade20k_20220105_151706.log.json) |
| Segmenter Linear | ViT-S_16 | 512x512 | 160000 | 1.78 | 28.07 | 45.75 | 46.82 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segmenter/segmenter_vit-s_linear_8x1_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segmenter/segmenter_vit-s_linear_8x1_512x512_160k_ade20k/segmenter_vit-s_linear_8x1_512x512_160k_ade20k_20220105_151713-39658c46.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segmenter/segmenter_vit-s_linear_8x1_512x512_160k_ade20k/segmenter_vit-s_linear_8x1_512x512_160k_ade20k_20220105_151713.log.json) |
| Segmenter Mask | ViT-S_16 | 512x512 | 160000 | 2.03 | 24.80 | 46.19 | 47.85 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segmenter/segmenter_vit-s_mask_8x1_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segmenter/segmenter_vit-s_mask_8x1_512x512_160k_ade20k/segmenter_vit-s_mask_8x1_512x512_160k_ade20k_20220105_151706-511bb103.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segmenter/segmenter_vit-s_mask_8x1_512x512_160k_ade20k/segmenter_vit-s_mask_8x1_512x512_160k_ade20k_20220105_151706.log.json) |
| Segmenter Mask | ViT-B_16 | 512x512 | 160000 | 4.20 | 13.20 | 49.60 | 51.07 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segmenter/segmenter_vit-b_mask_8x1_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segmenter/segmenter_vit-b_mask_8x1_512x512_160k_ade20k/segmenter_vit-b_mask_8x1_512x512_160k_ade20k_20220105_151706-bc533b08.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segmenter/segmenter_vit-b_mask_8x1_512x512_160k_ade20k/segmenter_vit-b_mask_8x1_512x512_160k_ade20k_20220105_151706.log.json) |
| Segmenter Mask | ViT-L_16 | 640x640 | 160000 | 16.99 | 3.03 | 51.65 | 53.58 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segmenter/segmenter_vit-l_mask_8x1_640x640_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segmenter/segmenter_vit-l_mask_8x1_640x640_160k_ade20k/segmenter_vit-l_mask_8x1_640x640_160k_ade20k_20220614_024513-4783a347.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segmenter/segmenter_vit-l_mask_8x1_640x640_160k_ade20k/segmenter_vit-l_mask_8x1_640x640_160k_ade20k_20220614_024513.log.json) |
Note:
- This model performance is sensitive to the seed values used, please refer to the log file for the specific settings of the seed. If you choose a different seed, the results might differ from the table results.
| 9,362 | 117.518987 | 1,290 | md |
mmsegmentation | mmsegmentation-master/configs/segmenter/segmenter.yml | Collections:
- Name: Segmenter
Metadata:
Training Data:
- ADE20K
Paper:
URL: https://arxiv.org/abs/2105.05633
Title: 'Segmenter: Transformer for Semantic Segmentation'
README: configs/segmenter/README.md
Code:
URL: https://github.com/open-mmlab/mmsegmentation/blob/v0.21.0/mmseg/models/decode_heads/segmenter_mask_head.py#L15
Version: v0.21.0
Converted From:
Code: https://github.com/rstrudel/segmenter
Models:
- Name: segmenter_vit-t_mask_8x1_512x512_160k_ade20k
In Collection: Segmenter
Metadata:
backbone: ViT-T_16
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 35.74
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 1.21
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 39.99
mIoU(ms+flip): 40.83
Config: configs/segmenter/segmenter_vit-t_mask_8x1_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segmenter/segmenter_vit-t_mask_8x1_512x512_160k_ade20k/segmenter_vit-t_mask_8x1_512x512_160k_ade20k_20220105_151706-ffcf7509.pth
- Name: segmenter_vit-s_linear_8x1_512x512_160k_ade20k
In Collection: Segmenter
Metadata:
backbone: ViT-S_16
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 35.63
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 1.78
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 45.75
mIoU(ms+flip): 46.82
Config: configs/segmenter/segmenter_vit-s_linear_8x1_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segmenter/segmenter_vit-s_linear_8x1_512x512_160k_ade20k/segmenter_vit-s_linear_8x1_512x512_160k_ade20k_20220105_151713-39658c46.pth
- Name: segmenter_vit-s_mask_8x1_512x512_160k_ade20k
In Collection: Segmenter
Metadata:
backbone: ViT-S_16
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 40.32
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 2.03
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 46.19
mIoU(ms+flip): 47.85
Config: configs/segmenter/segmenter_vit-s_mask_8x1_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segmenter/segmenter_vit-s_mask_8x1_512x512_160k_ade20k/segmenter_vit-s_mask_8x1_512x512_160k_ade20k_20220105_151706-511bb103.pth
- Name: segmenter_vit-b_mask_8x1_512x512_160k_ade20k
In Collection: Segmenter
Metadata:
backbone: ViT-B_16
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 75.76
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 4.2
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 49.6
mIoU(ms+flip): 51.07
Config: configs/segmenter/segmenter_vit-b_mask_8x1_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segmenter/segmenter_vit-b_mask_8x1_512x512_160k_ade20k/segmenter_vit-b_mask_8x1_512x512_160k_ade20k_20220105_151706-bc533b08.pth
- Name: segmenter_vit-l_mask_8x1_640x640_160k_ade20k
In Collection: Segmenter
Metadata:
backbone: ViT-L_16
crop size: (640,640)
lr schd: 160000
inference time (ms/im):
- value: 330.03
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (640,640)
Training Memory (GB): 16.99
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 51.65
mIoU(ms+flip): 53.58
Config: configs/segmenter/segmenter_vit-l_mask_8x1_640x640_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segmenter/segmenter_vit-l_mask_8x1_640x640_160k_ade20k/segmenter_vit-l_mask_8x1_640x640_160k_ade20k_20220614_024513-4783a347.pth
| 4,132 | 31.801587 | 194 | yml |
mmsegmentation | mmsegmentation-master/configs/segmenter/segmenter_vit-b_mask_8x1_512x512_160k_ade20k.py | _base_ = [
'../_base_/models/segmenter_vit-b16_mask.py',
'../_base_/datasets/ade20k.py', '../_base_/default_runtime.py',
'../_base_/schedules/schedule_160k.py'
]
optimizer = dict(lr=0.001, weight_decay=0.0)
img_norm_cfg = dict(
mean=[127.5, 127.5, 127.5], std=[127.5, 127.5, 127.5], to_rgb=True)
crop_size = (512, 512)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadAnnotations', reduce_zero_label=True),
dict(type='Resize', img_scale=(2048, 512), ratio_range=(0.5, 2.0)),
dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
dict(type='RandomFlip', prob=0.5),
dict(type='PhotoMetricDistortion'),
dict(type='Normalize', **img_norm_cfg),
dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img', 'gt_semantic_seg'])
]
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=(2048, 512),
# img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
flip=False,
transforms=[
dict(type='Resize', keep_ratio=True),
dict(type='RandomFlip'),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img'])
])
]
data = dict(
# num_gpus: 8 -> batch_size: 8
samples_per_gpu=1,
train=dict(pipeline=train_pipeline),
val=dict(pipeline=test_pipeline),
test=dict(pipeline=test_pipeline))
| 1,546 | 34.159091 | 71 | py |
mmsegmentation | mmsegmentation-master/configs/segmenter/segmenter_vit-l_mask_8x1_640x640_160k_ade20k.py | _base_ = [
'../_base_/models/segmenter_vit-b16_mask.py',
'../_base_/datasets/ade20k_640x640.py', '../_base_/default_runtime.py',
'../_base_/schedules/schedule_160k.py'
]
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segmenter/vit_large_p16_384_20220308-d4efb41d.pth' # noqa
model = dict(
pretrained=checkpoint,
backbone=dict(
type='VisionTransformer',
img_size=(640, 640),
embed_dims=1024,
num_layers=24,
num_heads=16),
decode_head=dict(
type='SegmenterMaskTransformerHead',
in_channels=1024,
channels=1024,
num_heads=16,
embed_dims=1024),
test_cfg=dict(mode='slide', crop_size=(640, 640), stride=(608, 608)))
optimizer = dict(lr=0.001, weight_decay=0.0)
img_norm_cfg = dict(
mean=[127.5, 127.5, 127.5], std=[127.5, 127.5, 127.5], to_rgb=True)
crop_size = (640, 640)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadAnnotations', reduce_zero_label=True),
dict(type='Resize', img_scale=(2560, 640), ratio_range=(0.5, 2.0)),
dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
dict(type='RandomFlip', prob=0.5),
dict(type='PhotoMetricDistortion'),
dict(type='Normalize', **img_norm_cfg),
dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img', 'gt_semantic_seg'])
]
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=(2560, 640),
# img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
flip=False,
transforms=[
dict(type='Resize', keep_ratio=True),
dict(type='RandomFlip'),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img'])
])
]
data = dict(
# num_gpus: 8 -> batch_size: 8
samples_per_gpu=1,
train=dict(pipeline=train_pipeline),
val=dict(pipeline=test_pipeline),
test=dict(pipeline=test_pipeline))
| 2,121 | 33.225806 | 132 | py |
mmsegmentation | mmsegmentation-master/configs/segmenter/segmenter_vit-s_linear_8x1_512x512_160k_ade20k.py | _base_ = './segmenter_vit-s_mask_8x1_512x512_160k_ade20k.py'
model = dict(
decode_head=dict(
_delete_=True,
type='FCNHead',
in_channels=384,
channels=384,
num_convs=0,
dropout_ratio=0.0,
concat_input=False,
num_classes=150,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)))
| 394 | 25.333333 | 74 | py |
mmsegmentation | mmsegmentation-master/configs/segmenter/segmenter_vit-s_mask_8x1_512x512_160k_ade20k.py | _base_ = [
'../_base_/models/segmenter_vit-b16_mask.py',
'../_base_/datasets/ade20k.py', '../_base_/default_runtime.py',
'../_base_/schedules/schedule_160k.py'
]
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segmenter/vit_small_p16_384_20220308-410f6037.pth' # noqa
backbone_norm_cfg = dict(type='LN', eps=1e-6, requires_grad=True)
model = dict(
pretrained=checkpoint,
backbone=dict(
img_size=(512, 512),
embed_dims=384,
num_heads=6,
),
decode_head=dict(
type='SegmenterMaskTransformerHead',
in_channels=384,
channels=384,
num_classes=150,
num_layers=2,
num_heads=6,
embed_dims=384,
dropout_ratio=0.0,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)))
optimizer = dict(lr=0.001, weight_decay=0.0)
img_norm_cfg = dict(
mean=[127.5, 127.5, 127.5], std=[127.5, 127.5, 127.5], to_rgb=True)
crop_size = (512, 512)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadAnnotations', reduce_zero_label=True),
dict(type='Resize', img_scale=(2048, 512), ratio_range=(0.5, 2.0)),
dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
dict(type='RandomFlip', prob=0.5),
dict(type='PhotoMetricDistortion'),
dict(type='Normalize', **img_norm_cfg),
dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img', 'gt_semantic_seg'])
]
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=(2048, 512),
# img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
flip=False,
transforms=[
dict(type='Resize', keep_ratio=True),
dict(type='RandomFlip'),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img'])
])
]
data = dict(
# num_gpus: 8 -> batch_size: 8
samples_per_gpu=1,
train=dict(pipeline=train_pipeline),
val=dict(pipeline=test_pipeline),
test=dict(pipeline=test_pipeline))
| 2,223 | 32.19403 | 132 | py |
mmsegmentation | mmsegmentation-master/configs/segmenter/segmenter_vit-t_mask_8x1_512x512_160k_ade20k.py | _base_ = [
'../_base_/models/segmenter_vit-b16_mask.py',
'../_base_/datasets/ade20k.py', '../_base_/default_runtime.py',
'../_base_/schedules/schedule_160k.py'
]
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segmenter/vit_tiny_p16_384_20220308-cce8c795.pth' # noqa
model = dict(
pretrained=checkpoint,
backbone=dict(embed_dims=192, num_heads=3),
decode_head=dict(
type='SegmenterMaskTransformerHead',
in_channels=192,
channels=192,
num_heads=3,
embed_dims=192))
optimizer = dict(lr=0.001, weight_decay=0.0)
img_norm_cfg = dict(
mean=[127.5, 127.5, 127.5], std=[127.5, 127.5, 127.5], to_rgb=True)
crop_size = (512, 512)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadAnnotations', reduce_zero_label=True),
dict(type='Resize', img_scale=(2048, 512), ratio_range=(0.5, 2.0)),
dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
dict(type='RandomFlip', prob=0.5),
dict(type='PhotoMetricDistortion'),
dict(type='Normalize', **img_norm_cfg),
dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img', 'gt_semantic_seg'])
]
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=(2048, 512),
# img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
flip=False,
transforms=[
dict(type='Resize', keep_ratio=True),
dict(type='RandomFlip'),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img'])
])
]
data = dict(
# num_gpus: 8 -> batch_size: 8
samples_per_gpu=1,
train=dict(pipeline=train_pipeline),
val=dict(pipeline=test_pipeline),
test=dict(pipeline=test_pipeline))
| 1,930 | 32.877193 | 131 | py |
mmsegmentation | mmsegmentation-master/configs/segnext/README.md | # SegNeXt
[SegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentation](https://arxiv.org/abs/2209.08575)
## Introduction
<!-- [ALGORITHM] -->
<a href="https://github.com/visual-attention-network/segnext">Official Repo</a>
<a href="https://github.com/open-mmlab/mmsegmentation/blob/v0.31.0/mmseg/models/backbones/mscan.py#L328">Code Snippet</a>
## Abstract
<!-- [ABSTRACT] -->
We present SegNeXt, a simple convolutional network architecture for semantic segmentation. Recent transformer-based models have dominated the field of semantic segmentation due to the efficiency of self-attention in encoding spatial information. In this paper, we show that convolutional attention is a more efficient and effective way to encode contextual information than the self-attention mechanism in transformers. By re-examining the characteristics owned by successful segmentation models, we discover several key components leading to the performance improvement of segmentation models. This motivates us to design a novel convolutional attention network that uses cheap convolutional operations. Without bells and whistles, our SegNeXt significantly improves the performance of previous state-of-the-art methods on popular benchmarks, including ADE20K, Cityscapes, COCO-Stuff, Pascal VOC, Pascal Context, and iSAID. Notably, SegNeXt outperforms EfficientNet-L2 w/ NAS-FPN and achieves 90.6% mIoU on the Pascal VOC 2012 test leaderboard using only 1/10 parameters of it. On average, SegNeXt achieves about 2.0% mIoU improvements compared to the state-of-the-art methods on the ADE20K datasets with the same or fewer computations. Code is available at [this https URL](https://github.com/uyzhang/JSeg) (Jittor) and [this https URL](https://github.com/Visual-Attention-Network/SegNeXt) (Pytorch).
<!-- [IMAGE] -->
<div align=center>
<img src="https://user-images.githubusercontent.com/24582831/215688018-5d4c8366-7793-4fdf-9397-960a09fac951.png" width="70%"/>
</div>
```bibtex
@article{guo2022segnext,
title={SegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentation},
author={Guo, Meng-Hao and Lu, Cheng-Ze and Hou, Qibin and Liu, Zhengning and Cheng, Ming-Ming and Hu, Shi-Min},
journal={arXiv preprint arXiv:2209.08575},
year={2022}
}
```
## Results and models
### ADE20K
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| ------- | -------- | --------- | ------- | -------- | -------------- | ----- | ------------- | ------------------------------------------------------------------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| SegNeXt | MSCAN-T | 512x512 | 160000 | 17.88 | 52.38 | 41.50 | 42.59 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segnext/segnext_mscan-t_1x16_512x512_adamw_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segnext/segnext_mscan-t_1x16_512x512_adamw_160k_ade20k/segnext_mscan-t_1x16_512x512_adamw_160k_ade20k_20230210_140244-05bd8466.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segnext/segnext_mscan-t_1x16_512x512_adamw_160k_ade20k/segnext_mscan-t_1x16_512x512_adamw_160k_ade20k_20230210_140244.log.json) |
| SegNeXt | MSCAN-S | 512x512 | 160000 | 21.47 | 42.27 | 44.16 | 45.81 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segnext/segnext_mscan-s_1x16_512x512_adamw_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segnext/segnext_mscan-s_1x16_512x512_adamw_160k_ade20k/segnext_mscan-s_1x16_512x512_adamw_160k_ade20k_20230214_113014-43013668.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segnext/segnext_mscan-s_1x16_512x512_adamw_160k_ade20k/segnext_mscan-s_1x16_512x512_adamw_160k_ade20k_20230214_113014.log.json) |
| SegNeXt | MSCAN-B | 512x512 | 160000 | 31.03 | 35.15 | 48.03 | 49.68 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segnext/segnext_mscan-b_1x16_512x512_adamw_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segnext/segnext_mscan-b_1x16_512x512_adamw_160k_ade20k/segnext_mscan-b_1x16_512x512_adamw_160k_ade20k_20230209_172053-b6f6c70c.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segnext/segnext_mscan-b_1x16_512x512_adamw_160k_ade20k/segnext_mscan-b_1x16_512x512_adamw_160k_ade20k_20230209_172053.log.json) |
| SegNeXt | MSCAN-L | 512x512 | 160000 | 43.32 | 22.91 | 50.99 | 52.10 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/segnext/segnext_mscan-l_1x16_512x512_adamw_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/segnext/segnext_mscan-l_1x16_512x512_adamw_160k_ade20k/segnext_mscan-l_1x16_512x512_adamw_160k_ade20k_20230209_172055-19b14b63.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/segnext/segnext_mscan-l_1x16_512x512_adamw_160k_ade20k/segnext_mscan-l_1x16_512x512_adamw_160k_ade20k_20230209_172055.log.json) |
Note:
- When we integrated SegNeXt into MMSegmentation, we modified some layers' names to make them more precise and concise without changing the model architecture. Therefore, the keys of pre-trained weights are different from the [original weights](https://cloud.tsinghua.edu.cn/d/c15b25a6745946618462/), but don't worry about these changes. we have converted them and uploaded the checkpoints, you might find URL of pre-trained checkpoints in config files and can use them directly for training.
- The total batch size is 16. We trained for SegNeXt with a single GPU as the performance degrades significantly when using`SyncBN` (mainly in `OverlapPatchEmbed` modules of `MSCAN`) of PyTorch 1.9.
- There will be subtle differences when model testing as Non-negative Matrix Factorization (NMF) in `LightHamHead` will be initialized randomly. To control this randomness, please set the random seed when model testing. You can modify [`./tools/test.py`](https://github.com/open-mmlab/mmsegmentation/blob/master/tools/test.py) like:
```python
def main():
from mmseg.apis import set_random_seed
random_seed = xxx # set random seed recorded in training log
set_random_seed(random_seed, deterministic=False)
...
```
- This model performance is sensitive to the seed values used, please refer to the log file for the specific settings of the seed. If you choose a different seed, the results might differ from the table results. Take SegNeXt Large for example, its results range from 49.60 to 51.0.
| 7,509 | 120.129032 | 1,402 | md |
mmsegmentation | mmsegmentation-master/configs/segnext/segnext.yml | Collections:
- Name: SegNeXt
Metadata:
Training Data:
- ADE20K
Paper:
URL: https://arxiv.org/abs/2209.08575
Title: 'SegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentation'
README: configs/segnext/README.md
Code:
URL: https://github.com/open-mmlab/mmsegmentation/blob/v0.31.0/mmseg/models/backbones/mscan.py#L328
Version: v0.31.0
Converted From:
Code: https://github.com/visual-attention-network/segnext
Models:
- Name: segnext_mscan-t_1x16_512x512_adamw_160k_ade20k
In Collection: SegNeXt
Metadata:
backbone: MSCAN-T
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 19.09
hardware: A100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 17.88
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 41.5
mIoU(ms+flip): 42.59
Config: configs/segnext/segnext_mscan-t_1x16_512x512_adamw_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segnext/segnext_mscan-t_1x16_512x512_adamw_160k_ade20k/segnext_mscan-t_1x16_512x512_adamw_160k_ade20k_20230210_140244-05bd8466.pth
- Name: segnext_mscan-s_1x16_512x512_adamw_160k_ade20k
In Collection: SegNeXt
Metadata:
backbone: MSCAN-S
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 23.66
hardware: A100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 21.47
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 44.16
mIoU(ms+flip): 45.81
Config: configs/segnext/segnext_mscan-s_1x16_512x512_adamw_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segnext/segnext_mscan-s_1x16_512x512_adamw_160k_ade20k/segnext_mscan-s_1x16_512x512_adamw_160k_ade20k_20230214_113014-43013668.pth
- Name: segnext_mscan-b_1x16_512x512_adamw_160k_ade20k
In Collection: SegNeXt
Metadata:
backbone: MSCAN-B
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 28.45
hardware: A100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 31.03
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 48.03
mIoU(ms+flip): 49.68
Config: configs/segnext/segnext_mscan-b_1x16_512x512_adamw_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segnext/segnext_mscan-b_1x16_512x512_adamw_160k_ade20k/segnext_mscan-b_1x16_512x512_adamw_160k_ade20k_20230209_172053-b6f6c70c.pth
- Name: segnext_mscan-l_1x16_512x512_adamw_160k_ade20k
In Collection: SegNeXt
Metadata:
backbone: MSCAN-L
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 43.65
hardware: A100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 43.32
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 50.99
mIoU(ms+flip): 52.1
Config: configs/segnext/segnext_mscan-l_1x16_512x512_adamw_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/segnext/segnext_mscan-l_1x16_512x512_adamw_160k_ade20k/segnext_mscan-l_1x16_512x512_adamw_160k_ade20k_20230209_172055-19b14b63.pth
| 3,417 | 31.865385 | 192 | yml |
mmsegmentation | mmsegmentation-master/configs/segnext/segnext_mscan-b_1x16_512x512_adamw_160k_ade20k.py | _base_ = './segnext_mscan-t_1x16_512x512_adamw_160k_ade20k.py'
# model settings
checkpoint_file = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segnext/mscan_b_20230227-3ab7d230.pth' # noqa
ham_norm_cfg = dict(type='GN', num_groups=32, requires_grad=True)
model = dict(
type='EncoderDecoder',
backbone=dict(
embed_dims=[64, 128, 320, 512],
depths=[3, 3, 12, 3],
init_cfg=dict(type='Pretrained', checkpoint=checkpoint_file),
drop_path_rate=0.1,
norm_cfg=dict(type='BN', requires_grad=True)),
decode_head=dict(
type='LightHamHead',
in_channels=[128, 320, 512],
in_index=[1, 2, 3],
channels=512,
ham_channels=512,
dropout_ratio=0.1,
num_classes=150,
norm_cfg=ham_norm_cfg,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
# model training and testing settings
train_cfg=dict(),
test_cfg=dict(mode='whole'))
| 1,029 | 35.785714 | 125 | py |
mmsegmentation | mmsegmentation-master/configs/segnext/segnext_mscan-l_1x16_512x512_adamw_160k_ade20k.py | _base_ = './segnext_mscan-t_1x16_512x512_adamw_160k_ade20k.py'
# model settings
checkpoint_file = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segnext/mscan_l_20230227-cef260d4.pth' # noqa
ham_norm_cfg = dict(type='GN', num_groups=32, requires_grad=True)
model = dict(
type='EncoderDecoder',
backbone=dict(
embed_dims=[64, 128, 320, 512],
depths=[3, 5, 27, 3],
init_cfg=dict(type='Pretrained', checkpoint=checkpoint_file),
drop_path_rate=0.3,
norm_cfg=dict(type='BN', requires_grad=True)),
decode_head=dict(
type='LightHamHead',
in_channels=[128, 320, 512],
in_index=[1, 2, 3],
channels=1024,
ham_channels=1024,
dropout_ratio=0.1,
num_classes=150,
norm_cfg=ham_norm_cfg,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
# model training and testing settings
train_cfg=dict(),
test_cfg=dict(mode='whole'))
| 1,031 | 35.857143 | 125 | py |
mmsegmentation | mmsegmentation-master/configs/segnext/segnext_mscan-s_1x16_512x512_adamw_160k_ade20k.py | _base_ = './segnext_mscan-t_1x16_512x512_adamw_160k_ade20k.py'
# model settings
checkpoint_file = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segnext/mscan_s_20230227-f33ccdf2.pth' # noqa
ham_norm_cfg = dict(type='GN', num_groups=32, requires_grad=True)
model = dict(
type='EncoderDecoder',
backbone=dict(
embed_dims=[64, 128, 320, 512],
depths=[2, 2, 4, 2],
init_cfg=dict(type='Pretrained', checkpoint=checkpoint_file),
norm_cfg=dict(type='BN', requires_grad=True)),
decode_head=dict(
type='LightHamHead',
in_channels=[128, 320, 512],
in_index=[1, 2, 3],
channels=256,
ham_channels=256,
ham_kwargs=dict(MD_R=16),
dropout_ratio=0.1,
num_classes=150,
norm_cfg=ham_norm_cfg,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0)),
# model training and testing settings
train_cfg=dict(),
test_cfg=dict(mode='whole'))
| 1,034 | 35.964286 | 125 | py |
mmsegmentation | mmsegmentation-master/configs/segnext/segnext_mscan-t_1x16_512x512_adamw_160k_ade20k.py | _base_ = [
'../_base_/default_runtime.py', '../_base_/schedules/schedule_160k.py'
]
# model settings
checkpoint_file = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/segnext/mscan_t_20230227-119e8c9f.pth' # noqa
ham_norm_cfg = dict(type='GN', num_groups=32, requires_grad=True)
model = dict(
type='EncoderDecoder',
pretrained=None,
backbone=dict(
type='MSCAN',
init_cfg=dict(type='Pretrained', checkpoint=checkpoint_file),
embed_dims=[32, 64, 160, 256],
mlp_ratios=[8, 8, 4, 4],
drop_rate=0.0,
drop_path_rate=0.1,
depths=[3, 3, 5, 2],
attention_kernel_sizes=[5, [1, 7], [1, 11], [1, 21]],
attention_kernel_paddings=[2, [0, 3], [0, 5], [0, 10]],
act_cfg=dict(type='GELU'),
norm_cfg=dict(type='BN', requires_grad=True)),
decode_head=dict(
type='LightHamHead',
in_channels=[64, 160, 256],
in_index=[1, 2, 3],
channels=256,
ham_channels=256,
dropout_ratio=0.1,
num_classes=150,
norm_cfg=ham_norm_cfg,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=1.0),
ham_kwargs=dict(
MD_S=1,
MD_R=16,
train_steps=6,
eval_steps=7,
inv_t=100,
rand_init=True)),
# model training and testing settings
train_cfg=dict(),
test_cfg=dict(mode='whole'))
# dataset settings
dataset_type = 'ADE20KDataset'
data_root = 'data/ade/ADEChallengeData2016'
img_norm_cfg = dict(
mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
crop_size = (512, 512)
train_pipeline = [
dict(type='LoadImageFromFile'),
dict(type='LoadAnnotations', reduce_zero_label=True),
dict(type='Resize', img_scale=(2048, 512), ratio_range=(0.5, 2.0)),
dict(type='RandomCrop', crop_size=crop_size, cat_max_ratio=0.75),
dict(type='RandomFlip', prob=0.5),
dict(type='PhotoMetricDistortion'),
dict(type='Normalize', **img_norm_cfg),
dict(type='Pad', size=crop_size, pad_val=0, seg_pad_val=255),
dict(type='DefaultFormatBundle'),
dict(type='Collect', keys=['img', 'gt_semantic_seg']),
]
test_pipeline = [
dict(type='LoadImageFromFile'),
dict(
type='MultiScaleFlipAug',
img_scale=(2048, 512),
# img_ratios=[0.5, 0.75, 1.0, 1.25, 1.5, 1.75],
flip=False,
transforms=[
dict(type='Resize', keep_ratio=True),
dict(type='ResizeToMultiple', size_divisor=32),
dict(type='RandomFlip'),
dict(type='Normalize', **img_norm_cfg),
dict(type='ImageToTensor', keys=['img']),
dict(type='Collect', keys=['img']),
])
]
data = dict(
samples_per_gpu=16,
workers_per_gpu=4,
train=dict(
type='RepeatDataset',
times=50,
dataset=dict(
type=dataset_type,
data_root=data_root,
img_dir='images/training',
ann_dir='annotations/training',
pipeline=train_pipeline)),
val=dict(
type=dataset_type,
data_root=data_root,
img_dir='images/validation',
ann_dir='annotations/validation',
pipeline=test_pipeline),
test=dict(
type=dataset_type,
data_root=data_root,
img_dir='images/validation',
ann_dir='annotations/validation',
pipeline=test_pipeline))
# optimizer
optimizer = dict(
_delete_=True,
type='AdamW',
lr=0.00006,
betas=(0.9, 0.999),
weight_decay=0.01,
paramwise_cfg=dict(
custom_keys={
'pos_block': dict(decay_mult=0.),
'norm': dict(decay_mult=0.),
'head': dict(lr_mult=10.)
}))
lr_config = dict(
_delete_=True,
policy='poly',
warmup='linear',
warmup_iters=1500,
warmup_ratio=1e-6,
power=1.0,
min_lr=0.0,
by_epoch=False)
| 3,968 | 30.251969 | 125 | py |
mmsegmentation | mmsegmentation-master/configs/sem_fpn/README.md | # Semantic FPN
[Panoptic Feature Pyramid Networks](https://arxiv.org/abs/1901.02446)
## Introduction
<!-- [ALGORITHM] -->
<a href="https://github.com/facebookresearch/detectron2">Official Repo</a>
<a href="https://github.com/open-mmlab/mmsegmentation/blob/v0.17.0/mmseg/models/decode_heads/fpn_head.py#L12">Code Snippet</a>
## Abstract
<!-- [ABSTRACT] -->
The recently introduced panoptic segmentation task has renewed our community's interest in unifying the tasks of instance segmentation (for thing classes) and semantic segmentation (for stuff classes). However, current state-of-the-art methods for this joint task use separate and dissimilar networks for instance and semantic segmentation, without performing any shared computation. In this work, we aim to unify these methods at the architectural level, designing a single network for both tasks. Our approach is to endow Mask R-CNN, a popular instance segmentation method, with a semantic segmentation branch using a shared Feature Pyramid Network (FPN) backbone. Surprisingly, this simple baseline not only remains effective for instance segmentation, but also yields a lightweight, top-performing method for semantic segmentation. In this work, we perform a detailed study of this minimally extended version of Mask R-CNN with FPN, which we refer to as Panoptic FPN, and show it is a robust and accurate baseline for both tasks. Given its effectiveness and conceptual simplicity, we hope our method can serve as a strong baseline and aid future research in panoptic segmentation.
<!-- [IMAGE] -->
<div align=center>
<img src="https://user-images.githubusercontent.com/24582831/142902694-03ed2131-9104-467b-ace1-c74c62fb7177.png" width="60%"/>
</div>
## Citation
```bibtex
@inproceedings{kirillov2019panoptic,
title={Panoptic feature pyramid networks},
author={Kirillov, Alexander and Girshick, Ross and He, Kaiming and Doll{\'a}r, Piotr},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={6399--6408},
year={2019}
}
```
## Results and models
### Cityscapes
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| ------ | -------- | --------- | ------: | -------: | -------------- | ----: | ------------- | ---------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| FPN | R-50 | 512x1024 | 80000 | 2.8 | 13.54 | 74.52 | 76.08 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/sem_fpn/fpn_r50_512x1024_80k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/sem_fpn/fpn_r50_512x1024_80k_cityscapes/fpn_r50_512x1024_80k_cityscapes_20200717_021437-94018a0d.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/sem_fpn/fpn_r50_512x1024_80k_cityscapes/fpn_r50_512x1024_80k_cityscapes-20200717_021437.log.json) |
| FPN | R-101 | 512x1024 | 80000 | 3.9 | 10.29 | 75.80 | 77.40 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/sem_fpn/fpn_r101_512x1024_80k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/sem_fpn/fpn_r101_512x1024_80k_cityscapes/fpn_r101_512x1024_80k_cityscapes_20200717_012416-c5800d4c.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/sem_fpn/fpn_r101_512x1024_80k_cityscapes/fpn_r101_512x1024_80k_cityscapes-20200717_012416.log.json) |
### ADE20K
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| ------ | -------- | --------- | ------: | -------: | -------------- | ----: | ------------- | ------------------------------------------------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| FPN | R-50 | 512x512 | 160000 | 4.9 | 55.77 | 37.49 | 39.09 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/sem_fpn/fpn_r50_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/sem_fpn/fpn_r50_512x512_160k_ade20k/fpn_r50_512x512_160k_ade20k_20200718_131734-5b5a6ab9.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/sem_fpn/fpn_r50_512x512_160k_ade20k/fpn_r50_512x512_160k_ade20k-20200718_131734.log.json) |
| FPN | R-101 | 512x512 | 160000 | 5.9 | 40.58 | 39.35 | 40.72 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/sem_fpn/fpn_r101_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/sem_fpn/fpn_r101_512x512_160k_ade20k/fpn_r101_512x512_160k_ade20k_20200718_131734-306b5004.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/sem_fpn/fpn_r101_512x512_160k_ade20k/fpn_r101_512x512_160k_ade20k-20200718_131734.log.json) |
| 6,364 | 121.403846 | 1,184 | md |
mmsegmentation | mmsegmentation-master/configs/sem_fpn/fpn_r101_512x1024_80k_cityscapes.py | _base_ = './fpn_r50_512x1024_80k_cityscapes.py'
model = dict(pretrained='open-mmlab://resnet101_v1c', backbone=dict(depth=101))
| 128 | 42 | 79 | py |
mmsegmentation | mmsegmentation-master/configs/sem_fpn/fpn_r101_512x512_160k_ade20k.py | _base_ = './fpn_r50_512x512_160k_ade20k.py'
model = dict(pretrained='open-mmlab://resnet101_v1c', backbone=dict(depth=101))
| 124 | 40.666667 | 79 | py |
mmsegmentation | mmsegmentation-master/configs/sem_fpn/fpn_r50_512x1024_80k_cityscapes.py | _base_ = [
'../_base_/models/fpn_r50.py', '../_base_/datasets/cityscapes.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
| 158 | 30.8 | 73 | py |
mmsegmentation | mmsegmentation-master/configs/sem_fpn/fpn_r50_512x512_160k_ade20k.py | _base_ = [
'../_base_/models/fpn_r50.py', '../_base_/datasets/ade20k.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_160k.py'
]
model = dict(decode_head=dict(num_classes=150))
| 203 | 33 | 74 | py |
mmsegmentation | mmsegmentation-master/configs/sem_fpn/sem_fpn.yml | Collections:
- Name: FPN
Metadata:
Training Data:
- Cityscapes
- ADE20K
Paper:
URL: https://arxiv.org/abs/1901.02446
Title: Panoptic Feature Pyramid Networks
README: configs/sem_fpn/README.md
Code:
URL: https://github.com/open-mmlab/mmsegmentation/blob/v0.17.0/mmseg/models/decode_heads/fpn_head.py#L12
Version: v0.17.0
Converted From:
Code: https://github.com/facebookresearch/detectron2
Models:
- Name: fpn_r50_512x1024_80k_cityscapes
In Collection: FPN
Metadata:
backbone: R-50
crop size: (512,1024)
lr schd: 80000
inference time (ms/im):
- value: 73.86
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,1024)
Training Memory (GB): 2.8
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 74.52
mIoU(ms+flip): 76.08
Config: configs/sem_fpn/fpn_r50_512x1024_80k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/sem_fpn/fpn_r50_512x1024_80k_cityscapes/fpn_r50_512x1024_80k_cityscapes_20200717_021437-94018a0d.pth
- Name: fpn_r101_512x1024_80k_cityscapes
In Collection: FPN
Metadata:
backbone: R-101
crop size: (512,1024)
lr schd: 80000
inference time (ms/im):
- value: 97.18
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,1024)
Training Memory (GB): 3.9
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 75.8
mIoU(ms+flip): 77.4
Config: configs/sem_fpn/fpn_r101_512x1024_80k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/sem_fpn/fpn_r101_512x1024_80k_cityscapes/fpn_r101_512x1024_80k_cityscapes_20200717_012416-c5800d4c.pth
- Name: fpn_r50_512x512_160k_ade20k
In Collection: FPN
Metadata:
backbone: R-50
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 17.93
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 4.9
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 37.49
mIoU(ms+flip): 39.09
Config: configs/sem_fpn/fpn_r50_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/sem_fpn/fpn_r50_512x512_160k_ade20k/fpn_r50_512x512_160k_ade20k_20200718_131734-5b5a6ab9.pth
- Name: fpn_r101_512x512_160k_ade20k
In Collection: FPN
Metadata:
backbone: R-101
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 24.64
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 5.9
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 39.35
mIoU(ms+flip): 40.72
Config: configs/sem_fpn/fpn_r101_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/sem_fpn/fpn_r101_512x512_160k_ade20k/fpn_r101_512x512_160k_ade20k_20200718_131734-306b5004.pth
| 3,097 | 28.504762 | 164 | yml |
mmsegmentation | mmsegmentation-master/configs/setr/README.md | # SETR
[Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective with Transformers](https://arxiv.org/abs/2012.15840)
## Introduction
<!-- [ALGORITHM] -->
<a href="https://github.com/fudan-zvg/SETR">Official Repo</a>
<a href="https://github.com/open-mmlab/mmsegmentation/blob/v0.17.0/mmseg/models/decode_heads/setr_up_head.py#L11">Code Snippet</a>
## Abstract
<!-- [ABSTRACT] -->
Most recent semantic segmentation methods adopt a fully-convolutional network (FCN) with an encoder-decoder architecture. The encoder progressively reduces the spatial resolution and learns more abstract/semantic visual concepts with larger receptive fields. Since context modeling is critical for segmentation, the latest efforts have been focused on increasing the receptive field, through either dilated/atrous convolutions or inserting attention modules. However, the encoder-decoder based FCN architecture remains unchanged. In this paper, we aim to provide an alternative perspective by treating semantic segmentation as a sequence-to-sequence prediction task. Specifically, we deploy a pure transformer (ie, without convolution and resolution reduction) to encode an image as a sequence of patches. With the global context modeled in every layer of the transformer, this encoder can be combined with a simple decoder to provide a powerful segmentation model, termed SEgmentation TRansformer (SETR). Extensive experiments show that SETR achieves new state of the art on ADE20K (50.28% mIoU), Pascal Context (55.83% mIoU) and competitive results on Cityscapes. Particularly, we achieve the first position in the highly competitive ADE20K test server leaderboard on the day of submission.
<!-- [IMAGE] -->
<div align=center>
<img src="https://user-images.githubusercontent.com/24582831/142902777-ee2d34b7-a631-4fa7-ad68-118ff5716afe.png" width="80%"/>
</div>
```None
This head has two version head.
```
## Citation
```bibtex
@article{zheng2020rethinking,
title={Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective with Transformers},
author={Zheng, Sixiao and Lu, Jiachen and Zhao, Hengshuang and Zhu, Xiatian and Luo, Zekun and Wang, Yabiao and Fu, Yanwei and Feng, Jianfeng and Xiang, Tao and Torr, Philip HS and others},
journal={arXiv preprint arXiv:2012.15840},
year={2020}
}
```
## Usage
You can download the pretrain from [here](https://github.com/rwightman/pytorch-image-models/releases/download/v0.1-vitjx/jx_vit_large_p16_384-b3be5167.pth). Then you can convert its keys with the script `vit2mmseg.py` in the tools directory.
```shell
python tools/model_converters/vit2mmseg.py ${PRETRAIN_PATH} ${STORE_PATH}
```
E.g.
```shell
python tools/model_converters/vit2mmseg.py \
jx_vit_large_p16_384-b3be5167.pth pretrain/vit_large_p16.pth
```
This script convert the model from `PRETRAIN_PATH` and store the converted model in `STORE_PATH`.
## Results and models
### ADE20K
| Method | Backbone | Crop Size | Batch Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| ---------- | -------- | --------- | ---------- | ------- | -------- | -------------- | ----- | ------------: | --------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| SETR Naive | ViT-L | 512x512 | 16 | 160000 | 18.40 | 4.72 | 48.28 | 49.56 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/setr/setr_naive_512x512_160k_b16_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_naive_512x512_160k_b16_ade20k/setr_naive_512x512_160k_b16_ade20k_20210619_191258-061f24f5.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_naive_512x512_160k_b16_ade20k/setr_naive_512x512_160k_b16_ade20k_20210619_191258.log.json) |
| SETR PUP | ViT-L | 512x512 | 16 | 160000 | 19.54 | 4.50 | 48.24 | 49.99 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/setr/setr_pup_512x512_160k_b16_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_pup_512x512_160k_b16_ade20k/setr_pup_512x512_160k_b16_ade20k_20210619_191343-7e0ce826.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_pup_512x512_160k_b16_ade20k/setr_pup_512x512_160k_b16_ade20k_20210619_191343.log.json) |
| SETR MLA | ViT-L | 512x512 | 8 | 160000 | 10.96 | - | 47.34 | 49.05 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/setr/setr_mla_512x512_160k_b8_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_mla_512x512_160k_b8_ade20k/setr_mla_512x512_160k_b8_ade20k_20210619_191118-c6d21df0.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_mla_512x512_160k_b8_ade20k/setr_mla_512x512_160k_b8_ade20k_20210619_191118.log.json) |
| SETR MLA | ViT-L | 512x512 | 16 | 160000 | 17.30 | 5.25 | 47.39 | 49.37 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/setr/setr_mla_512x512_160k_b16_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_mla_512x512_160k_b16_ade20k/setr_mla_512x512_160k_b16_ade20k_20210619_191057-f9741de7.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_mla_512x512_160k_b16_ade20k/setr_mla_512x512_160k_b16_ade20k_20210619_191057.log.json) |
### Cityscapes
| Method | Backbone | Crop Size | Batch Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| ---------- | -------- | --------- | ---------- | ------- | -------- | -------------- | ----- | ------------: | ---------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| SETR Naive | ViT-L | 768x768 | 8 | 80000 | 24.06 | 0.39 | 78.10 | 80.22 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/setr/setr_vit-large_naive_8x1_768x768_80k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_naive_vit-large_8x1_768x768_80k_cityscapes/setr_naive_vit-large_8x1_768x768_80k_cityscapes_20211123_000505-20728e80.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_naive_vit-large_8x1_768x768_80k_cityscapes/setr_naive_vit-large_8x1_768x768_80k_cityscapes_20211123_000505.log.json) |
| SETR PUP | ViT-L | 768x768 | 8 | 80000 | 27.96 | 0.37 | 79.21 | 81.02 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/setr/setr_vit-large_pup_8x1_768x768_80k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_pup_vit-large_8x1_768x768_80k_cityscapes/setr_pup_vit-large_8x1_768x768_80k_cityscapes_20211122_155115-f6f37b8f.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_pup_vit-large_8x1_768x768_80k_cityscapes/setr_pup_vit-large_8x1_768x768_80k_cityscapes_20211122_155115.log.json) |
| SETR MLA | ViT-L | 768x768 | 8 | 80000 | 24.10 | 0.41 | 77.00 | 79.59 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/setr/setr_vit-large_mla_8x1_768x768_80k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_mla_vit-large_8x1_768x768_80k_cityscapes/setr_mla_vit-large_8x1_768x768_80k_cityscapes_20211119_101003-7f8dccbe.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_mla_vit-large_8x1_768x768_80k_cityscapes/setr_mla_vit-large_8x1_768x768_80k_cityscapes_20211119_101003.log.json) |
| 9,431 | 124.76 | 1,292 | md |
mmsegmentation | mmsegmentation-master/configs/setr/setr.yml | Collections:
- Name: SETR
Metadata:
Training Data:
- ADE20K
- Cityscapes
Paper:
URL: https://arxiv.org/abs/2012.15840
Title: Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective
with Transformers
README: configs/setr/README.md
Code:
URL: https://github.com/open-mmlab/mmsegmentation/blob/v0.17.0/mmseg/models/decode_heads/setr_up_head.py#L11
Version: v0.17.0
Converted From:
Code: https://github.com/fudan-zvg/SETR
Models:
- Name: setr_naive_512x512_160k_b16_ade20k
In Collection: SETR
Metadata:
backbone: ViT-L
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 211.86
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 18.4
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 48.28
mIoU(ms+flip): 49.56
Config: configs/setr/setr_naive_512x512_160k_b16_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_naive_512x512_160k_b16_ade20k/setr_naive_512x512_160k_b16_ade20k_20210619_191258-061f24f5.pth
- Name: setr_pup_512x512_160k_b16_ade20k
In Collection: SETR
Metadata:
backbone: ViT-L
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 222.22
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 19.54
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 48.24
mIoU(ms+flip): 49.99
Config: configs/setr/setr_pup_512x512_160k_b16_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_pup_512x512_160k_b16_ade20k/setr_pup_512x512_160k_b16_ade20k_20210619_191343-7e0ce826.pth
- Name: setr_mla_512x512_160k_b8_ade20k
In Collection: SETR
Metadata:
backbone: ViT-L
crop size: (512,512)
lr schd: 160000
Training Memory (GB): 10.96
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 47.34
mIoU(ms+flip): 49.05
Config: configs/setr/setr_mla_512x512_160k_b8_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_mla_512x512_160k_b8_ade20k/setr_mla_512x512_160k_b8_ade20k_20210619_191118-c6d21df0.pth
- Name: setr_mla_512x512_160k_b16_ade20k
In Collection: SETR
Metadata:
backbone: ViT-L
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 190.48
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 17.3
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 47.39
mIoU(ms+flip): 49.37
Config: configs/setr/setr_mla_512x512_160k_b16_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_mla_512x512_160k_b16_ade20k/setr_mla_512x512_160k_b16_ade20k_20210619_191057-f9741de7.pth
- Name: setr_vit-large_naive_8x1_768x768_80k_cityscapes
In Collection: SETR
Metadata:
backbone: ViT-L
crop size: (768,768)
lr schd: 80000
inference time (ms/im):
- value: 2564.1
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (768,768)
Training Memory (GB): 24.06
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 78.1
mIoU(ms+flip): 80.22
Config: configs/setr/setr_vit-large_naive_8x1_768x768_80k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_naive_vit-large_8x1_768x768_80k_cityscapes/setr_naive_vit-large_8x1_768x768_80k_cityscapes_20211123_000505-20728e80.pth
- Name: setr_vit-large_pup_8x1_768x768_80k_cityscapes
In Collection: SETR
Metadata:
backbone: ViT-L
crop size: (768,768)
lr schd: 80000
inference time (ms/im):
- value: 2702.7
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (768,768)
Training Memory (GB): 27.96
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 79.21
mIoU(ms+flip): 81.02
Config: configs/setr/setr_vit-large_pup_8x1_768x768_80k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_pup_vit-large_8x1_768x768_80k_cityscapes/setr_pup_vit-large_8x1_768x768_80k_cityscapes_20211122_155115-f6f37b8f.pth
- Name: setr_vit-large_mla_8x1_768x768_80k_cityscapes
In Collection: SETR
Metadata:
backbone: ViT-L
crop size: (768,768)
lr schd: 80000
inference time (ms/im):
- value: 2439.02
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (768,768)
Training Memory (GB): 24.1
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 77.0
mIoU(ms+flip): 79.59
Config: configs/setr/setr_vit-large_mla_8x1_768x768_80k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/setr/setr_mla_vit-large_8x1_768x768_80k_cityscapes/setr_mla_vit-large_8x1_768x768_80k_cityscapes_20211119_101003-7f8dccbe.pth
| 5,204 | 30.545455 | 191 | yml |
mmsegmentation | mmsegmentation-master/configs/setr/setr_mla_512x512_160k_b16_ade20k.py | _base_ = ['./setr_mla_512x512_160k_b8_ade20k.py']
# num_gpus: 8 -> batch_size: 16
data = dict(samples_per_gpu=2)
| 114 | 22 | 49 | py |
mmsegmentation | mmsegmentation-master/configs/setr/setr_mla_512x512_160k_b8_ade20k.py | _base_ = [
'../_base_/models/setr_mla.py', '../_base_/datasets/ade20k.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_160k.py'
]
norm_cfg = dict(type='SyncBN', requires_grad=True)
model = dict(
pretrained=None,
backbone=dict(
img_size=(512, 512),
drop_rate=0.,
init_cfg=dict(
type='Pretrained', checkpoint='pretrain/vit_large_p16.pth')),
decode_head=dict(num_classes=150),
auxiliary_head=[
dict(
type='FCNHead',
in_channels=256,
channels=256,
in_index=0,
dropout_ratio=0,
norm_cfg=norm_cfg,
act_cfg=dict(type='ReLU'),
num_convs=0,
kernel_size=1,
concat_input=False,
num_classes=150,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
dict(
type='FCNHead',
in_channels=256,
channels=256,
in_index=1,
dropout_ratio=0,
norm_cfg=norm_cfg,
act_cfg=dict(type='ReLU'),
num_convs=0,
kernel_size=1,
concat_input=False,
num_classes=150,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
dict(
type='FCNHead',
in_channels=256,
channels=256,
in_index=2,
dropout_ratio=0,
norm_cfg=norm_cfg,
act_cfg=dict(type='ReLU'),
num_convs=0,
kernel_size=1,
concat_input=False,
num_classes=150,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
dict(
type='FCNHead',
in_channels=256,
channels=256,
in_index=3,
dropout_ratio=0,
norm_cfg=norm_cfg,
act_cfg=dict(type='ReLU'),
num_convs=0,
kernel_size=1,
concat_input=False,
num_classes=150,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
],
test_cfg=dict(mode='slide', crop_size=(512, 512), stride=(341, 341)),
)
optimizer = dict(
lr=0.001,
weight_decay=0.0,
paramwise_cfg=dict(custom_keys={'head': dict(lr_mult=10.)}))
# num_gpus: 8 -> batch_size: 8
data = dict(samples_per_gpu=1)
| 2,635 | 29.651163 | 78 | py |
mmsegmentation | mmsegmentation-master/configs/setr/setr_naive_512x512_160k_b16_ade20k.py | _base_ = [
'../_base_/models/setr_naive.py', '../_base_/datasets/ade20k.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_160k.py'
]
norm_cfg = dict(type='SyncBN', requires_grad=True)
model = dict(
pretrained=None,
backbone=dict(
img_size=(512, 512),
drop_rate=0.,
init_cfg=dict(
type='Pretrained', checkpoint='pretrain/vit_large_p16.pth')),
decode_head=dict(num_classes=150),
auxiliary_head=[
dict(
type='SETRUPHead',
in_channels=1024,
channels=256,
in_index=0,
num_classes=150,
dropout_ratio=0,
norm_cfg=norm_cfg,
act_cfg=dict(type='ReLU'),
num_convs=2,
kernel_size=1,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
dict(
type='SETRUPHead',
in_channels=1024,
channels=256,
in_index=1,
num_classes=150,
dropout_ratio=0,
norm_cfg=norm_cfg,
act_cfg=dict(type='ReLU'),
num_convs=2,
kernel_size=1,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
dict(
type='SETRUPHead',
in_channels=1024,
channels=256,
in_index=2,
num_classes=150,
dropout_ratio=0,
norm_cfg=norm_cfg,
act_cfg=dict(type='ReLU'),
num_convs=2,
kernel_size=1,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4))
],
test_cfg=dict(mode='slide', crop_size=(512, 512), stride=(341, 341)),
)
optimizer = dict(
lr=0.01,
weight_decay=0.0,
paramwise_cfg=dict(custom_keys={'head': dict(lr_mult=10.)}))
# num_gpus: 8 -> batch_size: 16
data = dict(samples_per_gpu=2)
| 2,077 | 29.558824 | 78 | py |
mmsegmentation | mmsegmentation-master/configs/setr/setr_pup_512x512_160k_b16_ade20k.py | _base_ = [
'../_base_/models/setr_pup.py', '../_base_/datasets/ade20k.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_160k.py'
]
norm_cfg = dict(type='SyncBN', requires_grad=True)
model = dict(
pretrained=None,
backbone=dict(
img_size=(512, 512),
drop_rate=0.,
init_cfg=dict(
type='Pretrained', checkpoint='pretrain/vit_large_p16.pth')),
decode_head=dict(num_classes=150),
auxiliary_head=[
dict(
type='SETRUPHead',
in_channels=1024,
channels=256,
in_index=0,
num_classes=150,
dropout_ratio=0,
norm_cfg=norm_cfg,
act_cfg=dict(type='ReLU'),
num_convs=2,
kernel_size=3,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
dict(
type='SETRUPHead',
in_channels=1024,
channels=256,
in_index=1,
num_classes=150,
dropout_ratio=0,
norm_cfg=norm_cfg,
act_cfg=dict(type='ReLU'),
num_convs=2,
kernel_size=3,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
dict(
type='SETRUPHead',
in_channels=1024,
channels=256,
in_index=2,
num_classes=150,
dropout_ratio=0,
norm_cfg=norm_cfg,
act_cfg=dict(type='ReLU'),
num_convs=2,
kernel_size=3,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
],
test_cfg=dict(mode='slide', crop_size=(512, 512), stride=(341, 341)),
)
optimizer = dict(
lr=0.001,
weight_decay=0.0,
paramwise_cfg=dict(custom_keys={'head': dict(lr_mult=10.)}))
# num_gpus: 8 -> batch_size: 16
data = dict(samples_per_gpu=2)
| 2,077 | 29.558824 | 78 | py |
mmsegmentation | mmsegmentation-master/configs/setr/setr_vit-large_mla_8x1_768x768_80k_cityscapes.py | _base_ = [
'../_base_/models/setr_mla.py', '../_base_/datasets/cityscapes_768x768.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
model = dict(
pretrained=None,
backbone=dict(
drop_rate=0,
init_cfg=dict(
type='Pretrained', checkpoint='pretrain/vit_large_p16.pth')),
test_cfg=dict(mode='slide', crop_size=(768, 768), stride=(512, 512)))
optimizer = dict(
lr=0.002,
weight_decay=0.0,
paramwise_cfg=dict(custom_keys={'head': dict(lr_mult=10.)}))
data = dict(samples_per_gpu=1)
| 564 | 30.388889 | 79 | py |
mmsegmentation | mmsegmentation-master/configs/setr/setr_vit-large_naive_8x1_768x768_80k_cityscapes.py | _base_ = [
'../_base_/models/setr_naive.py',
'../_base_/datasets/cityscapes_768x768.py', '../_base_/default_runtime.py',
'../_base_/schedules/schedule_80k.py'
]
model = dict(
pretrained=None,
backbone=dict(
drop_rate=0.,
init_cfg=dict(
type='Pretrained', checkpoint='pretrain/vit_large_p16.pth')),
test_cfg=dict(mode='slide', crop_size=(768, 768), stride=(512, 512)))
optimizer = dict(
weight_decay=0.0,
paramwise_cfg=dict(custom_keys={'head': dict(lr_mult=10.)}))
data = dict(samples_per_gpu=1)
| 558 | 28.421053 | 79 | py |
mmsegmentation | mmsegmentation-master/configs/setr/setr_vit-large_pup_8x1_768x768_80k_cityscapes.py | _base_ = [
'../_base_/models/setr_pup.py', '../_base_/datasets/cityscapes_768x768.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
norm_cfg = dict(type='SyncBN', requires_grad=True)
crop_size = (768, 768)
model = dict(
pretrained=None,
backbone=dict(
drop_rate=0.,
init_cfg=dict(
type='Pretrained', checkpoint='pretrain/vit_large_p16.pth')),
auxiliary_head=[
dict(
type='SETRUPHead',
in_channels=1024,
channels=256,
in_index=0,
num_classes=19,
dropout_ratio=0,
norm_cfg=norm_cfg,
num_convs=2,
up_scale=4,
kernel_size=3,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
dict(
type='SETRUPHead',
in_channels=1024,
channels=256,
in_index=1,
num_classes=19,
dropout_ratio=0,
norm_cfg=norm_cfg,
num_convs=2,
up_scale=4,
kernel_size=3,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4)),
dict(
type='SETRUPHead',
in_channels=1024,
channels=256,
in_index=2,
num_classes=19,
dropout_ratio=0,
norm_cfg=norm_cfg,
num_convs=2,
up_scale=4,
kernel_size=3,
align_corners=False,
loss_decode=dict(
type='CrossEntropyLoss', use_sigmoid=False, loss_weight=0.4))
],
test_cfg=dict(mode='slide', crop_size=crop_size, stride=(512, 512)))
optimizer = dict(
weight_decay=0.0,
paramwise_cfg=dict(custom_keys={'head': dict(lr_mult=10.)}))
data = dict(samples_per_gpu=1)
| 1,946 | 28.953846 | 79 | py |
mmsegmentation | mmsegmentation-master/configs/stdc/README.md | # STDC
[Rethinking BiSeNet For Real-time Semantic Segmentation](https://arxiv.org/abs/2104.13188)
## Introduction
<!-- [ALGORITHM] -->
<a href="https://github.com/MichaelFan01/STDC-Seg">Official Repo</a>
<a href="https://github.com/open-mmlab/mmsegmentation/blob/v0.20.0/mmseg/models/backbones/stdc.py#L394">Code Snippet</a>
## Abstract
<!-- [ABSTRACT] -->
BiSeNet has been proved to be a popular two-stream network for real-time segmentation. However, its principle of adding an extra path to encode spatial information is time-consuming, and the backbones borrowed from pretrained tasks, e.g., image classification, may be inefficient for image segmentation due to the deficiency of task-specific design. To handle these problems, we propose a novel and efficient structure named Short-Term Dense Concatenate network (STDC network) by removing structure redundancy. Specifically, we gradually reduce the dimension of feature maps and use the aggregation of them for image representation, which forms the basic module of STDC network. In the decoder, we propose a Detail Aggregation module by integrating the learning of spatial information into low-level layers in single-stream manner. Finally, the low-level features and deep features are fused to predict the final segmentation results. Extensive experiments on Cityscapes and CamVid dataset demonstrate the effectiveness of our method by achieving promising trade-off between segmentation accuracy and inference speed. On Cityscapes, we achieve 71.9% mIoU on the test set with a speed of 250.4 FPS on NVIDIA GTX 1080Ti, which is 45.2% faster than the latest methods, and achieve 76.8% mIoU with 97.0 FPS while inferring on higher resolution images.
<!-- [IMAGE] -->
<div align=center>
<img src="https://user-images.githubusercontent.com/24582831/143640374-d0709587-edb2-4821-bb60-340035f6ad8f.png" width="60%"/>
</div>
## Citation
```bibtex
@inproceedings{fan2021rethinking,
title={Rethinking BiSeNet For Real-time Semantic Segmentation},
author={Fan, Mingyuan and Lai, Shenqi and Huang, Junshi and Wei, Xiaoming and Chai, Zhenhua and Luo, Junfeng and Wei, Xiaolin},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={9716--9725},
year={2021}
}
```
## Usage
We have provided [ImageNet Pretrained STDCNet Weights](https://drive.google.com/drive/folders/1wROFwRt8qWHD4jSo8Zu1gp1d6oYJ3ns1) models converted from [official repo](https://github.com/MichaelFan01/STDC-Seg).
If you want to convert keys on your own to use official repositories' pre-trained models, we also provide a script [`stdc2mmseg.py`](../../tools/model_converters/stdc2mmseg.py) in the tools directory to convert the key of models from [the official repo](https://github.com/MichaelFan01/STDC-Seg) to MMSegmentation style.
```shell
python tools/model_converters/stdc2mmseg.py ${PRETRAIN_PATH} ${STORE_PATH} ${STDC_TYPE}
```
E.g.
```shell
python tools/model_converters/stdc2mmseg.py ./STDCNet813M_73.91.tar ./pretrained/stdc1.pth STDC1
python tools/model_converters/stdc2mmseg.py ./STDCNet1446_76.47.tar ./pretrained/stdc2.pth STDC2
```
This script convert model from `PRETRAIN_PATH` and store the converted model in `STORE_PATH`.
## Results and models
### Cityscapes
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| -------------------- | -------- | --------- | ------: | -------- | -------------- | ----: | ------------- | ------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| STDC 1 (No Pretrain) | STDC1 | 512x1024 | 80000 | 7.15 | 23.06 | 71.82 | 73.89 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/stdc/stdc1_512x1024_80k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/stdc/stdc1_512x1024_80k_cityscapes/stdc1_512x1024_80k_cityscapes_20220224_073048-74e6920a.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/stdc/stdc1_512x1024_80k_cityscapes/stdc1_512x1024_80k_cityscapes_20220224_073048.log.json) |
| STDC 1 | STDC1 | 512x1024 | 80000 | - | - | 74.94 | 76.97 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/stdc/stdc1_in1k-pre_512x1024_80k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/stdc/stdc1_in1k-pre_512x1024_80k_cityscapes/stdc1_in1k-pre_512x1024_80k_cityscapes_20220224_141648-3d4c2981.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/stdc/stdc1_in1k-pre_512x1024_80k_cityscapes/stdc1_in1k-pre_512x1024_80k_cityscapes_20220224_141648.log.json) |
| STDC 2 (No Pretrain) | STDC2 | 512x1024 | 80000 | 8.27 | 23.71 | 73.15 | 76.13 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/stdc/stdc2_512x1024_80k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/stdc/stdc2_512x1024_80k_cityscapes/stdc2_512x1024_80k_cityscapes_20220222_132015-fb1e3a1a.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/stdc/stdc2_512x1024_80k_cityscapes/stdc2_512x1024_80k_cityscapes_20220222_132015.log.json) |
| STDC 2 | STDC2 | 512x1024 | 80000 | - | - | 76.67 | 78.67 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/stdc/stdc2_in1k-pre_512x1024_80k_cityscapes.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/stdc/stdc2_in1k-pre_512x1024_80k_cityscapes/stdc2_in1k-pre_512x1024_80k_cityscapes_20220224_073048-1f8f0f6c.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/stdc/stdc2_in1k-pre_512x1024_80k_cityscapes/stdc2_in1k-pre_512x1024_80k_cityscapes_20220224_073048.log.json) |
Note:
- For STDC on Cityscapes dataset, default setting is 4 GPUs with 12 samples per GPU in training.
- `No Pretrain` means the model is trained from scratch.
- The FPS is for reference only. The environment is also different from paper setting, whose input size is `512x1024` and `768x1536`, i.e., 50% and 75% of our input size, respectively and using TensorRT.
- The parameter `fusion_kernel` in `STDCHead` is not learnable. In official repo, `find_unused_parameters=True` is set [here](https://github.com/MichaelFan01/STDC-Seg/blob/59ff37fbd693b99972c76fcefe97caa14aeb619f/train.py#L220). You may check it by printing model parameters of original repo on your own.
| 7,406 | 99.094595 | 1,347 | md |
mmsegmentation | mmsegmentation-master/configs/stdc/stdc.yml | Collections:
- Name: STDC
Metadata:
Training Data:
- Cityscapes
Paper:
URL: https://arxiv.org/abs/2104.13188
Title: Rethinking BiSeNet For Real-time Semantic Segmentation
README: configs/stdc/README.md
Code:
URL: https://github.com/open-mmlab/mmsegmentation/blob/v0.20.0/mmseg/models/backbones/stdc.py#L394
Version: v0.20.0
Converted From:
Code: https://github.com/MichaelFan01/STDC-Seg
Models:
- Name: stdc1_512x1024_80k_cityscapes
In Collection: STDC
Metadata:
backbone: STDC1
crop size: (512,1024)
lr schd: 80000
inference time (ms/im):
- value: 43.37
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,1024)
Training Memory (GB): 7.15
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 71.82
mIoU(ms+flip): 73.89
Config: configs/stdc/stdc1_512x1024_80k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/stdc/stdc1_512x1024_80k_cityscapes/stdc1_512x1024_80k_cityscapes_20220224_073048-74e6920a.pth
- Name: stdc1_in1k-pre_512x1024_80k_cityscapes
In Collection: STDC
Metadata:
backbone: STDC1
crop size: (512,1024)
lr schd: 80000
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 74.94
mIoU(ms+flip): 76.97
Config: configs/stdc/stdc1_in1k-pre_512x1024_80k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/stdc/stdc1_in1k-pre_512x1024_80k_cityscapes/stdc1_in1k-pre_512x1024_80k_cityscapes_20220224_141648-3d4c2981.pth
- Name: stdc2_512x1024_80k_cityscapes
In Collection: STDC
Metadata:
backbone: STDC2
crop size: (512,1024)
lr schd: 80000
inference time (ms/im):
- value: 42.18
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,1024)
Training Memory (GB): 8.27
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 73.15
mIoU(ms+flip): 76.13
Config: configs/stdc/stdc2_512x1024_80k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/stdc/stdc2_512x1024_80k_cityscapes/stdc2_512x1024_80k_cityscapes_20220222_132015-fb1e3a1a.pth
- Name: stdc2_in1k-pre_512x1024_80k_cityscapes
In Collection: STDC
Metadata:
backbone: STDC2
crop size: (512,1024)
lr schd: 80000
Results:
- Task: Semantic Segmentation
Dataset: Cityscapes
Metrics:
mIoU: 76.67
mIoU(ms+flip): 78.67
Config: configs/stdc/stdc2_in1k-pre_512x1024_80k_cityscapes.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/stdc/stdc2_in1k-pre_512x1024_80k_cityscapes/stdc2_in1k-pre_512x1024_80k_cityscapes_20220224_073048-1f8f0f6c.pth
| 2,777 | 30.568182 | 173 | yml |
mmsegmentation | mmsegmentation-master/configs/stdc/stdc1_512x1024_80k_cityscapes.py | _base_ = [
'../_base_/models/stdc.py', '../_base_/datasets/cityscapes.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
lr_config = dict(warmup='linear', warmup_iters=1000)
data = dict(
samples_per_gpu=12,
workers_per_gpu=4,
)
| 270 | 26.1 | 73 | py |
mmsegmentation | mmsegmentation-master/configs/stdc/stdc1_in1k-pre_512x1024_80k_cityscapes.py | checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/stdc/stdc1_20220308-5368626c.pth' # noqa
_base_ = './stdc1_512x1024_80k_cityscapes.py'
model = dict(
backbone=dict(
backbone_cfg=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint))))
| 293 | 41 | 115 | py |
mmsegmentation | mmsegmentation-master/configs/stdc/stdc2_512x1024_80k_cityscapes.py | _base_ = './stdc1_512x1024_80k_cityscapes.py'
model = dict(backbone=dict(backbone_cfg=dict(stdc_type='STDCNet2')))
| 115 | 37.666667 | 68 | py |
mmsegmentation | mmsegmentation-master/configs/stdc/stdc2_in1k-pre_512x1024_80k_cityscapes.py | checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/stdc/stdc2_20220308-7dbd9127.pth' # noqa
_base_ = './stdc2_512x1024_80k_cityscapes.py'
model = dict(
backbone=dict(
backbone_cfg=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint))))
| 293 | 41 | 115 | py |
mmsegmentation | mmsegmentation-master/configs/swin/README.md | # Swin Transformer
[Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://arxiv.org/abs/2103.14030)
## Introduction
<!-- [BACKBONE] -->
<a href="https://github.com/microsoft/Swin-Transformer">Official Repo</a>
<a href="https://github.com/open-mmlab/mmsegmentation/blob/v0.17.0/mmseg/models/backbones/swin.py#L524">Code Snippet</a>
## Abstract
<!-- [ABSTRACT] -->
This paper presents a new vision Transformer, called Swin Transformer, that capably serves as a general-purpose backbone for computer vision. Challenges in adapting Transformer from language to vision arise from differences between the two domains, such as large variations in the scale of visual entities and the high resolution of pixels in images compared to words in text. To address these differences, we propose a hierarchical Transformer whose representation is computed with Shifted windows. The shifted windowing scheme brings greater efficiency by limiting self-attention computation to non-overlapping local windows while also allowing for cross-window connection. This hierarchical architecture has the flexibility to model at various scales and has linear computational complexity with respect to image size. These qualities of Swin Transformer make it compatible with a broad range of vision tasks, including image classification (87.3 top-1 accuracy on ImageNet-1K) and dense prediction tasks such as object detection (58.7 box AP and 51.1 mask AP on COCO test-dev) and semantic segmentation (53.5 mIoU on ADE20K val). Its performance surpasses the previous state-of-the-art by a large margin of +2.7 box AP and +2.6 mask AP on COCO, and +3.2 mIoU on ADE20K, demonstrating the potential of Transformer-based models as vision backbones. The hierarchical design and the shifted window approach also prove beneficial for all-MLP architectures. The code and models are publicly available at [this https URL](https://github.com/microsoft/Swin-Transformer).
<!-- [IMAGE] -->
<div align=center>
<img src="https://user-images.githubusercontent.com/24582831/142902882-3fb9014c-11b6-47e9-aa14-500dfe7cbb1c.png" width="80%"/>
</div>
## Citation
```bibtex
@article{liu2021Swin,
title={Swin Transformer: Hierarchical Vision Transformer using Shifted Windows},
author={Liu, Ze and Lin, Yutong and Cao, Yue and Hu, Han and Wei, Yixuan and Zhang, Zheng and Lin, Stephen and Guo, Baining},
journal={arXiv preprint arXiv:2103.14030},
year={2021}
}
```
## Usage
We have provided pretrained models converted from [official repo](https://github.com/microsoft/Swin-Transformer)．
If you want to convert keys on your own to use official repositories' pre-trained models, we also provide a script [`swin2mmseg.py`](../../tools/model_converters/swin2mmseg.py) in the tools directory to convert the key of models from [the official repo](https://github.com/SwinTransformer/Swin-Transformer-Semantic-Segmentation) to MMSegmentation style.
```shell
python tools/model_converters/swin2mmseg.py ${PRETRAIN_PATH} ${STORE_PATH}
```
E.g.
```shell
python tools/model_converters/swin2mmseg.py https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window7_224.pth pretrain/swin_base_patch4_window7_224.pth
```
This script convert model from `PRETRAIN_PATH` and store the converted model in `STORE_PATH`.
In our default setting, pretrained models and their corresponding [original models](https://github.com/microsoft/Swin-Transforme) models could be defined below:
| pretrained models | original models |
| ----------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------- |
| pretrain/swin_tiny_patch4_window7_224.pth | [swin_tiny_patch4_window7_224.pth](https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_tiny_patch4_window7_224.pth) |
| pretrain/swin_small_patch4_window7_224.pth | [swin_small_patch4_window7_224.pth](https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_small_patch4_window7_224.pth) |
| pretrain/swin_base_patch4_window7_224.pth | [swin_base_patch4_window7_224.pth](https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window7_224.pth) |
| pretrain/swin_base_patch4_window7_224_22k.pth | [swin_base_patch4_window7_224_22k.pth](https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window7_224_22k.pth) |
| pretrain/swin_base_patch4_window12_384.pth | [swin_base_patch4_window12_384.pth](https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window12_384.pth) |
| pretrain/swin_base_patch4_window12_384_22k.pth | [swin_base_patch4_window12_384_22k.pth](https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_base_patch4_window12_384_22k.pth) |
| pretrain/swin_large_patch4_window7_224_22k.pth | [swin_large_patch4_window7_224_22k.pth](https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_large_patch4_window7_224_22k.pth) |
| pretrain/swin_large_patch4_window12_384_22k.pth | [swin_large_patch4_window12_384_22k.pth](https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_large_patch4_window12_384_22k.pth) |
## Results and models
### ADE20K
| Method | Backbone | Crop Size | pretrain | pretrain img size | Batch Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| ------- | -------- | --------- | ------------ | ----------------- | ---------- | ------- | -------- | -------------- | ----- | ------------: | -------------------------------------------------------------------------------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| UPerNet | Swin-T | 512x512 | ImageNet-1K | 224x224 | 16 | 160000 | 5.02 | 21.06 | 44.41 | 45.79 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/swin/upernet_swin_tiny_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_tiny_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K/upernet_swin_tiny_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K_20210531_112542-e380ad3e.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_tiny_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K/upernet_swin_tiny_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K_20210531_112542.log.json) |
| UPerNet | Swin-S | 512x512 | ImageNet-1K | 224x224 | 16 | 160000 | 6.17 | 14.72 | 47.72 | 49.24 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/swin/upernet_swin_small_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_small_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K/upernet_swin_small_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K_20210526_192015-ee2fff1c.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_small_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K/upernet_swin_small_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K_20210526_192015.log.json) |
| UPerNet | Swin-B | 512x512 | ImageNet-1K | 224x224 | 16 | 160000 | 7.61 | 12.65 | 47.99 | 49.57 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/swin/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K_20210526_192340-593b0e13.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K_20210526_192340.log.json) |
| UPerNet | Swin-B | 512x512 | ImageNet-22K | 224x224 | 16 | 160000 | - | - | 50.13 | 51.9 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/swin/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_22K.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_22K/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_22K_20210526_211650-762e2178.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_22K/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_22K_20210526_211650.log.json) |
| UPerNet | Swin-B | 512x512 | ImageNet-1K | 384x384 | 16 | 160000 | 8.52 | 12.10 | 48.35 | 49.65 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/swin/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_1K.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_1K/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_1K_20210531_132020-05b22ea4.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_1K/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_1K_20210531_132020.log.json) |
| UPerNet | Swin-B | 512x512 | ImageNet-22K | 384x384 | 16 | 160000 | - | - | 50.76 | 52.4 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/swin/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_22K.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_22K/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_22K_20210531_125459-429057bf.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_22K/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_22K_20210531_125459.log.json) |
| UPerNet | Swin-L | 512x512 | ImageNet-22K | 224x224 | 16 | 160000 | 10.98 | 8.23 | 51.17 | 52.99 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/swin/upernet_swin_large_patch4_window7_512x512_pretrain_224x224_22K_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_large_patch4_window7_512x512_pretrain_224x224_22K_160k_ade20k/upernet_swin_large_patch4_window7_512x512_pretrain_224x224_22K_160k_ade20k_20220318_015320-48d180dd.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_large_patch4_window7_512x512_pretrain_224x224_22K_160k_ade20k/upernet_swin_large_patch4_window7_512x512_pretrain_224x224_22K_160k_ade20k_20220318_015320.log.json) |
| UPerNet | Swin-L | 512x512 | ImageNet-22K | 384x384 | 16 | 160000 | 12.42 | 7.57 | 52.25 | 54.12 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/swin/upernet_swin_large_patch4_window12_512x512_pretrain_384x384_22K_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_large_patch4_window12_512x512_pretrain_384x384_22K_160k_ade20k/upernet_swin_large_patch4_window12_512x512_pretrain_384x384_22K_160k_ade20k_20220318_091743-9ba68901.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_large_patch4_window12_512x512_pretrain_384x384_22K_160k_ade20k/upernet_swin_large_patch4_window12_512x512_pretrain_384x384_22K_160k_ade20k_20220318_091743.log.json) |
| 13,529 | 166.037037 | 1,566 | md |
mmsegmentation | mmsegmentation-master/configs/swin/swin.yml | Models:
- Name: upernet_swin_tiny_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K
In Collection: UPerNet
Metadata:
backbone: Swin-T
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 47.48
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 5.02
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 44.41
mIoU(ms+flip): 45.79
Config: configs/swin/upernet_swin_tiny_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_tiny_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K/upernet_swin_tiny_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K_20210531_112542-e380ad3e.pth
- Name: upernet_swin_small_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K
In Collection: UPerNet
Metadata:
backbone: Swin-S
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 67.93
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 6.17
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 47.72
mIoU(ms+flip): 49.24
Config: configs/swin/upernet_swin_small_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_small_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K/upernet_swin_small_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K_20210526_192015-ee2fff1c.pth
- Name: upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K
In Collection: UPerNet
Metadata:
backbone: Swin-B
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 79.05
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 7.61
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 47.99
mIoU(ms+flip): 49.57
Config: configs/swin/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K_20210526_192340-593b0e13.pth
- Name: upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_22K
In Collection: UPerNet
Metadata:
backbone: Swin-B
crop size: (512,512)
lr schd: 160000
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 50.13
mIoU(ms+flip): 51.9
Config: configs/swin/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_22K.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_22K/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_22K_20210526_211650-762e2178.pth
- Name: upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_1K
In Collection: UPerNet
Metadata:
backbone: Swin-B
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 82.64
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 8.52
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 48.35
mIoU(ms+flip): 49.65
Config: configs/swin/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_1K.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_1K/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_1K_20210531_132020-05b22ea4.pth
- Name: upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_22K
In Collection: UPerNet
Metadata:
backbone: Swin-B
crop size: (512,512)
lr schd: 160000
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 50.76
mIoU(ms+flip): 52.4
Config: configs/swin/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_22K.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_22K/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_22K_20210531_125459-429057bf.pth
- Name: upernet_swin_large_patch4_window7_512x512_pretrain_224x224_22K_160k_ade20k
In Collection: UPerNet
Metadata:
backbone: Swin-L
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 121.51
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 10.98
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 51.17
mIoU(ms+flip): 52.99
Config: configs/swin/upernet_swin_large_patch4_window7_512x512_pretrain_224x224_22K_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_large_patch4_window7_512x512_pretrain_224x224_22K_160k_ade20k/upernet_swin_large_patch4_window7_512x512_pretrain_224x224_22K_160k_ade20k_20220318_015320-48d180dd.pth
- Name: upernet_swin_large_patch4_window12_512x512_pretrain_384x384_22K_160k_ade20k
In Collection: UPerNet
Metadata:
backbone: Swin-L
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 132.1
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 12.42
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 52.25
mIoU(ms+flip): 54.12
Config: configs/swin/upernet_swin_large_patch4_window12_512x512_pretrain_384x384_22K_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/swin/upernet_swin_large_patch4_window12_512x512_pretrain_384x384_22K_160k_ade20k/upernet_swin_large_patch4_window12_512x512_pretrain_384x384_22K_160k_ade20k_20220318_091743-9ba68901.pth
| 6,339 | 38.135802 | 247 | yml |
mmsegmentation | mmsegmentation-master/configs/swin/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_1K.py | _base_ = [
'upernet_swin_tiny_patch4_window7_512x512_160k_ade20k_'
'pretrain_224x224_1K.py'
]
checkpoint_file = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/swin/swin_base_patch4_window12_384_20220317-55b0104a.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint_file),
pretrain_img_size=384,
embed_dims=128,
depths=[2, 2, 18, 2],
num_heads=[4, 8, 16, 32],
window_size=12),
decode_head=dict(in_channels=[128, 256, 512, 1024], num_classes=150),
auxiliary_head=dict(in_channels=512, num_classes=150))
| 627 | 38.25 | 144 | py |
mmsegmentation | mmsegmentation-master/configs/swin/upernet_swin_base_patch4_window12_512x512_160k_ade20k_pretrain_384x384_22K.py | _base_ = [
'./upernet_swin_base_patch4_window12_512x512_160k_ade20k_'
'pretrain_384x384_1K.py'
]
checkpoint_file = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/swin/swin_base_patch4_window12_384_22k_20220317-e5c09f74.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint_file)))
| 358 | 38.888889 | 148 | py |
mmsegmentation | mmsegmentation-master/configs/swin/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K.py | _base_ = [
'./upernet_swin_tiny_patch4_window7_512x512_160k_ade20k_'
'pretrain_224x224_1K.py'
]
checkpoint_file = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/swin/swin_base_patch4_window7_224_20220317-e9b98025.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint_file),
embed_dims=128,
depths=[2, 2, 18, 2],
num_heads=[4, 8, 16, 32]),
decode_head=dict(in_channels=[128, 256, 512, 1024], num_classes=150),
auxiliary_head=dict(in_channels=512, num_classes=150))
| 573 | 40 | 143 | py |
mmsegmentation | mmsegmentation-master/configs/swin/upernet_swin_base_patch4_window7_512x512_160k_ade20k_pretrain_224x224_22K.py | _base_ = [
'./upernet_swin_base_patch4_window7_512x512_160k_ade20k_'
'pretrain_224x224_1K.py'
]
checkpoint_file = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/swin/swin_base_patch4_window7_224_22k_20220317-4f79f7c0.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint_file)))
| 356 | 38.666667 | 147 | py |
mmsegmentation | mmsegmentation-master/configs/swin/upernet_swin_large_patch4_window12_512x512_pretrain_384x384_22K_160k_ade20k.py | _base_ = [
'upernet_swin_large_patch4_window7_512x512_'
'pretrain_224x224_22K_160k_ade20k.py'
]
checkpoint_file = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/swin/swin_large_patch4_window12_384_22k_20220412-6580f57d.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint_file),
pretrain_img_size=384,
window_size=12))
| 413 | 36.636364 | 149 | py |
mmsegmentation | mmsegmentation-master/configs/swin/upernet_swin_large_patch4_window7_512x512_pretrain_224x224_22K_160k_ade20k.py | _base_ = [
'upernet_swin_tiny_patch4_window7_512x512_160k_ade20k_'
'pretrain_224x224_1K.py'
]
checkpoint_file = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/swin/swin_large_patch4_window7_224_22k_20220412-aeecf2aa.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint_file),
pretrain_img_size=224,
embed_dims=192,
depths=[2, 2, 18, 2],
num_heads=[6, 12, 24, 48],
window_size=7),
decode_head=dict(in_channels=[192, 384, 768, 1536], num_classes=150),
auxiliary_head=dict(in_channels=768, num_classes=150))
| 631 | 38.5 | 148 | py |
mmsegmentation | mmsegmentation-master/configs/swin/upernet_swin_small_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K.py | _base_ = [
'./upernet_swin_tiny_patch4_window7_512x512_160k_ade20k_'
'pretrain_224x224_1K.py'
]
checkpoint_file = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/swin/swin_small_patch4_window7_224_20220317-7ba6d6dd.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint_file),
depths=[2, 2, 18, 2]),
decode_head=dict(in_channels=[96, 192, 384, 768], num_classes=150),
auxiliary_head=dict(in_channels=384, num_classes=150))
| 514 | 41.916667 | 144 | py |
mmsegmentation | mmsegmentation-master/configs/swin/upernet_swin_tiny_patch4_window7_512x512_160k_ade20k_pretrain_224x224_1K.py | _base_ = [
'../_base_/models/upernet_swin.py', '../_base_/datasets/ade20k.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_160k.py'
]
checkpoint_file = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/swin/swin_tiny_patch4_window7_224_20220317-1cdeb081.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint_file),
embed_dims=96,
depths=[2, 2, 6, 2],
num_heads=[3, 6, 12, 24],
window_size=7,
use_abs_pos_embed=False,
drop_path_rate=0.3,
patch_norm=True),
decode_head=dict(in_channels=[96, 192, 384, 768], num_classes=150),
auxiliary_head=dict(in_channels=384, num_classes=150))
# AdamW optimizer, no weight decay for position embedding & layer norm
# in backbone
optimizer = dict(
_delete_=True,
type='AdamW',
lr=0.00006,
betas=(0.9, 0.999),
weight_decay=0.01,
paramwise_cfg=dict(
custom_keys={
'absolute_pos_embed': dict(decay_mult=0.),
'relative_position_bias_table': dict(decay_mult=0.),
'norm': dict(decay_mult=0.)
}))
lr_config = dict(
_delete_=True,
policy='poly',
warmup='linear',
warmup_iters=1500,
warmup_ratio=1e-6,
power=1.0,
min_lr=0.0,
by_epoch=False)
# By default, models are trained on 8 GPUs with 2 images per GPU
data = dict(samples_per_gpu=2)
| 1,428 | 30.065217 | 143 | py |
mmsegmentation | mmsegmentation-master/configs/twins/README.md | # Twins
[Twins: Revisiting the Design of Spatial Attention in Vision Transformers](https://arxiv.org/pdf/2104.13840.pdf)
## Introduction
<!-- [BACKBONE] -->
<a href = "https://github.com/Meituan-AutoML/Twins">Official Repo</a>
<a href="https://github.com/open-mmlab/mmsegmentation/blob/v0.20.0/mmseg/models/backbones/twins.py#L352">Code Snippet</a>
## Abstract
<!-- [ABSTRACT] -->
Very recently, a variety of vision transformer architectures for dense prediction tasks have been proposed and they show that the design of spatial attention is critical to their success in these tasks. In this work, we revisit the design of the spatial attention and demonstrate that a carefully-devised yet simple spatial attention mechanism performs favourably against the state-of-the-art schemes. As a result, we propose two vision transformer architectures, namely, Twins-PCPVT and Twins-SVT. Our proposed architectures are highly-efficient and easy to implement, only involving matrix multiplications that are highly optimized in modern deep learning frameworks. More importantly, the proposed architectures achieve excellent performance on a wide range of visual tasks, including image level classification as well as dense detection and segmentation. The simplicity and strong performance suggest that our proposed architectures may serve as stronger backbones for many vision tasks. Our code is released at [this https URL](https://github.com/Meituan-AutoML/Twins).
<!-- [IMAGE] -->
<div align=center>
<img src="https://user-images.githubusercontent.com/24582831/145021310-57826cf5-5e03-4c7c-9081-ffa744bdae27.png" width="80%"/>
</div>
## Citation
```bibtex
@article{chu2021twins,
title={Twins: Revisiting spatial attention design in vision transformers},
author={Chu, Xiangxiang and Tian, Zhi and Wang, Yuqing and Zhang, Bo and Ren, Haibing and Wei, Xiaolin and Xia, Huaxia and Shen, Chunhua},
journal={arXiv preprint arXiv:2104.13840},
year={2021}altgvt
}
```
## Usage
We have provided pretrained models converted from [official repo](https://github.com/Meituan-AutoML/Twins).
If you want to convert keys on your own to use official repositories' pre-trained models, we also provide a script [`twins2mmseg.py`](../../tools/model_converters/twins2mmseg.py) in the tools directory to convert the key of models from [the official repo](https://github.com/Meituan-AutoML/Twins) to MMSegmentation style.
```shell
python tools/model_converters/twins2mmseg.py ${PRETRAIN_PATH} ${STORE_PATH} ${MODEL_TYPE}
```
This script convert `pcpvt` or `svt` pretrained model from `PRETRAIN_PATH` and store the converted model in `STORE_PATH`.
For example,
```shell
python tools/model_converters/twins2mmseg.py ./alt_gvt_base.pth ./pretrained/alt_gvt_base.pth svt
```
## Results and models
### ADE20K
| Method | Backbone | Crop Size | Lr schd | Mem (GB) | Inf time (fps) | mIoU | mIoU(ms+flip) | config | download |
| ------------------- | -------- | --------- | ------- | -------- | -------------- | ----- | ------------- | ------------------------------------------------------------------------------------------------------------------------------------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Twins-FPN | PCPVT-S | 512x512 | 80000 | 6.60 | 27.15 | 43.26 | 44.11 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/twins/twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k_20211201_204132-41acd132.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k_20211201_204132.log.json) |
| Twins-UPerNet | PCPVT-S | 512x512 | 160000 | 9.67 | 14.24 | 46.04 | 46.92 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/twins/twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k/twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k_20211201_233537-8e99c07a.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k/twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k_20211201_233537.log.json) |
| Twins-FPN | PCPVT-B | 512x512 | 80000 | 8.41 | 19.67 | 45.66 | 46.48 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/twins/twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k_20211130_141019-d396db72.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k_20211130_141019.log.json) |
| Twins-UPerNet (8x2) | PCPVT-B | 512x512 | 160000 | 6.46 | 12.04 | 47.91 | 48.64 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/twins/twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k/twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k_20211130_141020-02094ea5.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k/twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k_20211130_141020.log.json) |
| Twins-FPN | PCPVT-L | 512x512 | 80000 | 10.78 | 14.32 | 45.94 | 46.70 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/twins/twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k_20211201_105226-bc6d61dc.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k_20211201_105226.log.json) |
| Twins-UPerNet (8x2) | PCPVT-L | 512x512 | 160000 | 7.82 | 10.70 | 49.35 | 50.08 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/twins/twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k/twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k_20211201_075053-c6095c07.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k/twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k_20211201_075053.log.json) |
| Twins-FPN | SVT-S | 512x512 | 80000 | 5.80 | 29.79 | 44.47 | 45.42 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/twins/twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k_20211130_141006-0a0d3317.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k_20211130_141006.log.json) |
| Twins-UPerNet (8x2) | SVT-S | 512x512 | 160000 | 4.93 | 15.09 | 46.08 | 46.96 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/twins/twins_svt-s_uperhead_8x2_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-s_uperhead_8x2_512x512_160k_ade20k/twins_svt-s_uperhead_8x2_512x512_160k_ade20k_20211130_141005-e48a2d94.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-s_uperhead_8x2_512x512_160k_ade20k/twins_svt-s_uperhead_8x2_512x512_160k_ade20k_20211130_141005.log.json) |
| Twins-FPN | SVT-B | 512x512 | 80000 | 8.75 | 21.10 | 46.77 | 47.47 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/twins/twins_svt-b_fpn_fpnhead_8x4_512x512_80k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-b_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_svt-b_fpn_fpnhead_8x4_512x512_80k_ade20k_20211201_113849-88b2907c.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-b_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_svt-b_fpn_fpnhead_8x4_512x512_80k_ade20k_20211201_113849.log.json) |
| Twins-UPerNet (8x2) | SVT-B | 512x512 | 160000 | 6.77 | 12.66 | 48.04 | 48.87 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/twins/twins_svt-b_uperhead_8x2_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-b_uperhead_8x2_512x512_160k_ade20k/twins_svt-b_uperhead_8x2_512x512_160k_ade20k_20211202_040826-0943a1f1.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-b_uperhead_8x2_512x512_160k_ade20k/twins_svt-b_uperhead_8x2_512x512_160k_ade20k_20211202_040826.log.json) |
| Twins-FPN | SVT-L | 512x512 | 80000 | 11.20 | 17.80 | 46.55 | 47.74 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/twins/twins_svt-l_fpn_fpnhead_8x4_512x512_80k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-l_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_svt-l_fpn_fpnhead_8x4_512x512_80k_ade20k_20211130_141005-1d59bee2.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-l_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_svt-l_fpn_fpnhead_8x4_512x512_80k_ade20k_20211130_141005.log.json) |
| Twins-UPerNet (8x2) | SVT-L | 512x512 | 160000 | 8.41 | 10.73 | 49.65 | 50.63 | [config](https://github.com/open-mmlab/mmsegmentation/blob/master/configs/twins/twins_svt-l_uperhead_8x2_512x512_160k_ade20k.py) | [model](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-l_uperhead_8x2_512x512_160k_ade20k/twins_svt-l_uperhead_8x2_512x512_160k_ade20k_20211130_141005-3e2cae61.pth) \| [log](https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-l_uperhead_8x2_512x512_160k_ade20k/twins_svt-l_uperhead_8x2_512x512_160k_ade20k_20211130_141005.log.json) |
Note:
- `8x2` means 8 GPUs with 2 samples per GPU in training. Default setting of Twins on ADE20K is 8 GPUs with 4 samples per GPU in training.
- `UPerNet` and `FPN` are decoder heads utilized in corresponding Twins model, which is `UPerHead` and `FPNHead`, respectively. Specifically, models in [official repo](https://github.com/Meituan-AutoML/Twins) all use `UPerHead`.
| 11,985 | 154.662338 | 1,075 | md |
mmsegmentation | mmsegmentation-master/configs/twins/twins.yml | Models:
- Name: twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k
In Collection: FPN
Metadata:
backbone: PCPVT-S
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 36.83
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 6.6
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 43.26
mIoU(ms+flip): 44.11
Config: configs/twins/twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k_20211201_204132-41acd132.pth
- Name: twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k
In Collection: UPerNet
Metadata:
backbone: PCPVT-S
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 70.22
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 9.67
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 46.04
mIoU(ms+flip): 46.92
Config: configs/twins/twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k/twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k_20211201_233537-8e99c07a.pth
- Name: twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k
In Collection: FPN
Metadata:
backbone: PCPVT-B
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 50.84
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 8.41
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 45.66
mIoU(ms+flip): 46.48
Config: configs/twins/twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k_20211130_141019-d396db72.pth
- Name: twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k
In Collection: UPerNet
Metadata:
backbone: PCPVT-B
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 83.06
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 6.46
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 47.91
mIoU(ms+flip): 48.64
Config: configs/twins/twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k/twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k_20211130_141020-02094ea5.pth
- Name: twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k
In Collection: FPN
Metadata:
backbone: PCPVT-L
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 69.83
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 10.78
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 45.94
mIoU(ms+flip): 46.7
Config: configs/twins/twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k_20211201_105226-bc6d61dc.pth
- Name: twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k
In Collection: UPerNet
Metadata:
backbone: PCPVT-L
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 93.46
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 7.82
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 49.35
mIoU(ms+flip): 50.08
Config: configs/twins/twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k/twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k_20211201_075053-c6095c07.pth
- Name: twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k
In Collection: FPN
Metadata:
backbone: SVT-S
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 33.57
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 5.8
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 44.47
mIoU(ms+flip): 45.42
Config: configs/twins/twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_svt-s_fpn_fpnhead_8x4_512x512_80k_ade20k_20211130_141006-0a0d3317.pth
- Name: twins_svt-s_uperhead_8x2_512x512_160k_ade20k
In Collection: UPerNet
Metadata:
backbone: SVT-S
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 66.27
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 4.93
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 46.08
mIoU(ms+flip): 46.96
Config: configs/twins/twins_svt-s_uperhead_8x2_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-s_uperhead_8x2_512x512_160k_ade20k/twins_svt-s_uperhead_8x2_512x512_160k_ade20k_20211130_141005-e48a2d94.pth
- Name: twins_svt-b_fpn_fpnhead_8x4_512x512_80k_ade20k
In Collection: FPN
Metadata:
backbone: SVT-B
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 47.39
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 8.75
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 46.77
mIoU(ms+flip): 47.47
Config: configs/twins/twins_svt-b_fpn_fpnhead_8x4_512x512_80k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-b_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_svt-b_fpn_fpnhead_8x4_512x512_80k_ade20k_20211201_113849-88b2907c.pth
- Name: twins_svt-b_uperhead_8x2_512x512_160k_ade20k
In Collection: UPerNet
Metadata:
backbone: SVT-B
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 78.99
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 6.77
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 48.04
mIoU(ms+flip): 48.87
Config: configs/twins/twins_svt-b_uperhead_8x2_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-b_uperhead_8x2_512x512_160k_ade20k/twins_svt-b_uperhead_8x2_512x512_160k_ade20k_20211202_040826-0943a1f1.pth
- Name: twins_svt-l_fpn_fpnhead_8x4_512x512_80k_ade20k
In Collection: FPN
Metadata:
backbone: SVT-L
crop size: (512,512)
lr schd: 80000
inference time (ms/im):
- value: 56.18
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 11.2
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 46.55
mIoU(ms+flip): 47.74
Config: configs/twins/twins_svt-l_fpn_fpnhead_8x4_512x512_80k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-l_fpn_fpnhead_8x4_512x512_80k_ade20k/twins_svt-l_fpn_fpnhead_8x4_512x512_80k_ade20k_20211130_141005-1d59bee2.pth
- Name: twins_svt-l_uperhead_8x2_512x512_160k_ade20k
In Collection: UPerNet
Metadata:
backbone: SVT-L
crop size: (512,512)
lr schd: 160000
inference time (ms/im):
- value: 93.2
hardware: V100
backend: PyTorch
batch size: 1
mode: FP32
resolution: (512,512)
Training Memory (GB): 8.41
Results:
- Task: Semantic Segmentation
Dataset: ADE20K
Metrics:
mIoU: 49.65
mIoU(ms+flip): 50.63
Config: configs/twins/twins_svt-l_uperhead_8x2_512x512_160k_ade20k.py
Weights: https://download.openmmlab.com/mmsegmentation/v0.5/twins/twins_svt-l_uperhead_8x2_512x512_160k_ade20k/twins_svt-l_uperhead_8x2_512x512_160k_ade20k_20211130_141005-3e2cae61.pth
| 8,747 | 31.887218 | 194 | yml |
mmsegmentation | mmsegmentation-master/configs/twins/twins_pcpvt-b_fpn_fpnhead_8x4_512x512_80k_ade20k.py | _base_ = ['./twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k.py']
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/pcpvt_base_20220308-0621964c.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
depths=[3, 4, 18, 3]), )
| 322 | 34.888889 | 121 | py |
mmsegmentation | mmsegmentation-master/configs/twins/twins_pcpvt-b_uperhead_8x2_512x512_160k_ade20k.py | _base_ = ['./twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k.py']
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/pcpvt_base_20220308-0621964c.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
depths=[3, 4, 18, 3],
drop_path_rate=0.3))
data = dict(samples_per_gpu=2, workers_per_gpu=2)
| 397 | 32.166667 | 121 | py |
mmsegmentation | mmsegmentation-master/configs/twins/twins_pcpvt-l_fpn_fpnhead_8x4_512x512_80k_ade20k.py | _base_ = ['./twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k.py']
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/pcpvt_large_20220308-37579dc6.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
depths=[3, 8, 27, 3]))
| 321 | 34.777778 | 122 | py |
mmsegmentation | mmsegmentation-master/configs/twins/twins_pcpvt-l_uperhead_8x2_512x512_160k_ade20k.py | _base_ = ['./twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k.py']
checkpoint = 'https://download.openmmlab.com/mmsegmentation/v0.5/pretrain/twins/pcpvt_large_20220308-37579dc6.pth' # noqa
model = dict(
backbone=dict(
init_cfg=dict(type='Pretrained', checkpoint=checkpoint),
depths=[3, 8, 27, 3],
drop_path_rate=0.3))
data = dict(samples_per_gpu=2, workers_per_gpu=2)
| 398 | 32.25 | 122 | py |
mmsegmentation | mmsegmentation-master/configs/twins/twins_pcpvt-s_fpn_fpnhead_8x4_512x512_80k_ade20k.py | _base_ = [
'../_base_/models/twins_pcpvt-s_fpn.py', '../_base_/datasets/ade20k.py',
'../_base_/default_runtime.py', '../_base_/schedules/schedule_80k.py'
]
optimizer = dict(_delete_=True, type='AdamW', lr=0.0001, weight_decay=0.0001)
| 243 | 33.857143 | 77 | py |
mmsegmentation | mmsegmentation-master/configs/twins/twins_pcpvt-s_uperhead_8x4_512x512_160k_ade20k.py | _base_ = [
'../_base_/models/twins_pcpvt-s_upernet.py',
'../_base_/datasets/ade20k.py', '../_base_/default_runtime.py',
'../_base_/schedules/schedule_160k.py'
]
optimizer = dict(
_delete_=True,
type='AdamW',
lr=0.00006,
betas=(0.9, 0.999),
weight_decay=0.01,
paramwise_cfg=dict(custom_keys={
'pos_block': dict(decay_mult=0.),
'norm': dict(decay_mult=0.)
}))
lr_config = dict(
_delete_=True,
policy='poly',
warmup='linear',
warmup_iters=1500,
warmup_ratio=1e-6,
power=1.0,
min_lr=0.0,
by_epoch=False)
| 590 | 20.888889 | 67 | py |
Subsets and Splits