Datasets:
metadata
dataset_info:
features:
- name: image
dtype: image
- name: depth
dtype: image
- name: label
dtype: image
splits:
- name: train
num_bytes: 152339736
num_examples: 795
- name: test
num_bytes: 124128749
num_examples: 654
download_size: 276518209
dataset_size: 276468485
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
task_categories:
- image-segmentation
NYUDepth v2
Easier version for semantic segmentation.
Comes from:
Indoor Segmentation and Support Inference from RGBD Images (ECCV 2012) (PDF) (Bib)