Update Superimposed-Masked-Dataset.py
Browse files- Superimposed-Masked-Dataset.py +26 -33
Superimposed-Masked-Dataset.py
CHANGED
@@ -1,13 +1,26 @@
|
|
1 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2 |
|
3 |
import os
|
4 |
-
import numpy as np
|
5 |
|
6 |
import datasets
|
7 |
from datasets.tasks import ImageClassification
|
8 |
|
9 |
from .classes import IMAGENET2012_CLASSES
|
10 |
-
from io import BytesIO
|
11 |
|
12 |
|
13 |
_CITATION = """\
|
@@ -28,11 +41,6 @@ _DATA_URL = {
|
|
28 |
]
|
29 |
}
|
30 |
|
31 |
-
_MASK_DATA_URL = {
|
32 |
-
"smd_masks": [
|
33 |
-
f"https://huggingface.co/datasets/ariellee/Superimposed-Masked-Dataset/resolve/main/SMD_masks.tar.gz"
|
34 |
-
]
|
35 |
-
}
|
36 |
|
37 |
class SMD(datasets.GeneratorBasedBuilder):
|
38 |
VERSION = datasets.Version("1.0.0")
|
@@ -47,7 +55,6 @@ class SMD(datasets.GeneratorBasedBuilder):
|
|
47 |
{
|
48 |
"image": datasets.Image(),
|
49 |
"label": datasets.ClassLabel(names=list(IMAGENET2012_CLASSES.values())),
|
50 |
-
"segmentation": datasets.Sequence(datasets.Array2D(shape=(None, None), dtype="float32"))
|
51 |
}
|
52 |
),
|
53 |
homepage=_HOMEPAGE,
|
@@ -57,40 +64,26 @@ class SMD(datasets.GeneratorBasedBuilder):
|
|
57 |
|
58 |
def _split_generators(self, dl_manager):
|
59 |
"""Returns SplitGenerators."""
|
60 |
-
archives = dl_manager.
|
61 |
-
|
62 |
-
|
63 |
return [
|
64 |
datasets.SplitGenerator(
|
65 |
-
name="SMD",
|
66 |
gen_kwargs={
|
67 |
-
"archives": archives["smd"],
|
68 |
-
"mask_archives": mask_archives["smd_masks"],
|
69 |
},
|
70 |
),
|
71 |
]
|
72 |
-
|
73 |
-
|
|
|
74 |
"""Yields examples."""
|
75 |
idx = 0
|
76 |
-
mask_files = {}
|
77 |
-
for mask_archive in mask_archives:
|
78 |
-
for path, file in dl_manager.iter_archive(mask_archive):
|
79 |
-
if path.endswith(".npy"):
|
80 |
-
mask_files[path] = np.load(BytesIO(file.read()))
|
81 |
-
|
82 |
for archive in archives:
|
83 |
-
for path, file in
|
84 |
if path.endswith(".png"):
|
85 |
synset_id = os.path.basename(os.path.dirname(path))
|
86 |
label = IMAGENET2012_CLASSES[synset_id]
|
87 |
-
|
88 |
-
mask_file_path = path.replace("_occluded.png", "_mask.npy")
|
89 |
-
segmentation_mask = mask_files.get(mask_file_path)
|
90 |
-
ex = {
|
91 |
-
"image": {"path": path, "bytes": file.read()},
|
92 |
-
"label": label,
|
93 |
-
"segmentation": segmentation_mask.tolist() # Convert numpy array to list
|
94 |
-
}
|
95 |
yield idx, ex
|
96 |
-
idx += 1
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
# Copyright 2022 the HuggingFace Datasets Authors.
|
3 |
+
#
|
4 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
5 |
+
# you may not use this file except in compliance with the License.
|
6 |
+
# You may obtain a copy of the License at
|
7 |
+
#
|
8 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
9 |
+
#
|
10 |
+
# Unless required by applicable law or agreed to in writing, software
|
11 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
12 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
13 |
+
# See the License for the specific language governing permissions and
|
14 |
+
# limitations under the License.
|
15 |
+
|
16 |
+
# This script was modified from the imagenet-1k HF dataset repo
|
17 |
|
18 |
import os
|
|
|
19 |
|
20 |
import datasets
|
21 |
from datasets.tasks import ImageClassification
|
22 |
|
23 |
from .classes import IMAGENET2012_CLASSES
|
|
|
24 |
|
25 |
|
26 |
_CITATION = """\
|
|
|
41 |
]
|
42 |
}
|
43 |
|
|
|
|
|
|
|
|
|
|
|
44 |
|
45 |
class SMD(datasets.GeneratorBasedBuilder):
|
46 |
VERSION = datasets.Version("1.0.0")
|
|
|
55 |
{
|
56 |
"image": datasets.Image(),
|
57 |
"label": datasets.ClassLabel(names=list(IMAGENET2012_CLASSES.values())),
|
|
|
58 |
}
|
59 |
),
|
60 |
homepage=_HOMEPAGE,
|
|
|
64 |
|
65 |
def _split_generators(self, dl_manager):
|
66 |
"""Returns SplitGenerators."""
|
67 |
+
archives = dl_manager.download(_DATA_URL)
|
68 |
+
|
|
|
69 |
return [
|
70 |
datasets.SplitGenerator(
|
71 |
+
name="SMD", # "SMD (occluded IN-1K val set)"
|
72 |
gen_kwargs={
|
73 |
+
"archives": [dl_manager.iter_archive(archive) for archive in archives["smd"]],
|
|
|
74 |
},
|
75 |
),
|
76 |
]
|
77 |
+
|
78 |
+
|
79 |
+
def _generate_examples(self, archives):
|
80 |
"""Yields examples."""
|
81 |
idx = 0
|
|
|
|
|
|
|
|
|
|
|
|
|
82 |
for archive in archives:
|
83 |
+
for path, file in archive:
|
84 |
if path.endswith(".png"):
|
85 |
synset_id = os.path.basename(os.path.dirname(path))
|
86 |
label = IMAGENET2012_CLASSES[synset_id]
|
87 |
+
ex = {"image": {"path": path, "bytes": file.read()}, "label": label}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
88 |
yield idx, ex
|
89 |
+
idx += 1
|