karolyartur commited on
Commit
b998535
1 Parent(s): b721f0a

Add loading script

Browse files
Files changed (1) hide show
  1. oe_dataset.py +147 -0
oe_dataset.py ADDED
@@ -0,0 +1,147 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #
2
+ # This file is part of the oe_dataset distribution (https://huggingface.co/datasets/ABC-iRobotics/oe_dataset).
3
+ # Copyright (c) 2023 ABC-iRobotics.
4
+ #
5
+ # This program is free software: you can redistribute it and/or modify
6
+ # it under the terms of the GNU General Public License as published by
7
+ # the Free Software Foundation, version 3.
8
+ #
9
+ # This program is distributed in the hope that it will be useful, but
10
+ # WITHOUT ANY WARRANTY; without even the implied warranty of
11
+ # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12
+ # General Public License for more details.
13
+ #
14
+ # You should have received a copy of the GNU General Public License
15
+ # along with this program. If not, see <http://www.gnu.org/licenses/>.
16
+ #
17
+ """OE dataset"""
18
+
19
+ from collections.abc import Sequence
20
+ from typing import Optional, Generator, Tuple, IO
21
+
22
+ import datasets
23
+
24
+
25
+ # ---- Constants ----
26
+
27
+ _CITATION = """\
28
+ @ARTICLE{10145828,
29
+ author={Károly, Artúr István and Tirczka, Sebestyén and Gao, Huijun and Rudas, Imre J. and Galambos, Péter},
30
+ journal={IEEE Transactions on Cybernetics},
31
+ title={Increasing the Robustness of Deep Learning Models for Object Segmentation: A Framework for Blending Automatically Annotated Real and Synthetic Data},
32
+ year={2023},
33
+ volume={},
34
+ number={},
35
+ pages={1-14},
36
+ doi={10.1109/TCYB.2023.3276485}}
37
+
38
+ """
39
+
40
+ _DESCRIPTION = """\
41
+ An instance segmentation dataset for robotic manipulation in a tabletop environment.
42
+ The dataset incorporates real and synthetic images for testing sim-to-real model transfer after fine-tuning.
43
+ """
44
+
45
+ _HOMEPAGE = "https://huggingface.co/ABC-iRobotics/oe_dataset"
46
+
47
+ _LICENSE = "GNU General Public License v3.0"
48
+
49
+ _LATEST_VERSIONS = {
50
+ "all": "1.0.0",
51
+ "real": "1.0.0",
52
+ "synthetic": "1.0.0",
53
+ "photoreal": "1.0.0",
54
+ "random": "1.0.0",
55
+ }
56
+
57
+
58
+
59
+ # ---- OE dataset Configs ----
60
+
61
+ class OEDatasetConfig(datasets.BuilderConfig):
62
+ """BuilderConfig for OE dataset."""
63
+
64
+ def __init__(self, name: str, imgs_urls: Sequence[str], masks_urls: Sequence[str], version: Optional[str] = None, **kwargs):
65
+ _version = _LATEST_VERSIONS[name] if version is None else version
66
+ _name = f"{name}_v{_version}"
67
+ super(OEDatasetConfig, self).__init__(version=datasets.Version(_version), name=_name, **kwargs)
68
+ self._imgs_urls = {"train": [url + "/train.tar.gz" for url in imgs_urls], "val": [url + "/val.tar.gz" for url in imgs_urls]}
69
+ self._masks_urls = {"train": [url + "/train.tar.gz" for url in masks_urls], "val": [url + "/val.tar.gz" for url in masks_urls]}
70
+
71
+ @property
72
+ def features(self):
73
+ return datasets.Features(
74
+ {
75
+ "image": datasets.Image(),
76
+ "mask": datasets.Image(),
77
+ }
78
+ )
79
+
80
+ @property
81
+ def supervised_keys(self):
82
+ return ("image", "mask")
83
+
84
+
85
+
86
+ # ---- OE dataset Loader ----
87
+
88
+ class OEDataset(datasets.GeneratorBasedBuilder):
89
+ """OE dataset."""
90
+
91
+ BUILDER_CONFIG_CLASS = OEDatasetConfig
92
+ BUILDER_CONFIGS = [
93
+ OEDatasetConfig(
94
+ name = "photoreal",
95
+ description = "Photorealistic synthetic images",
96
+ imgs_urls = ["https://huggingface.co/datasets/ABC-iRobotics/oe_dataset/synthetic/photoreal/imgs"],
97
+ masks_urls = ["https://huggingface.co/datasets/ABC-iRobotics/oe_dataset/synthetic/photoreal/masks"]
98
+ ),
99
+ ]
100
+
101
+ def _info(self):
102
+ return datasets.DatasetInfo(
103
+ description=_DESCRIPTION,
104
+ features=self.config.features,
105
+ supervised_keys=self.config.supervised_keys,
106
+ homepage=_HOMEPAGE,
107
+ license=_LICENSE,
108
+ citation=_CITATION,
109
+ version=self.config.version,
110
+ )
111
+
112
+ def _split_generators(self, dl_manager):
113
+ train_imgs_paths = dl_manager.download(self.config._imgs_urls["train"])
114
+ val_imgs_paths = dl_manager.download(self.config._imgs_urls["val"])
115
+
116
+ train_masks_paths = dl_manager.download(self.config._masks_urls["train"])
117
+ val_masks_paths = dl_manager.download(self.config._masks_urls["val"])
118
+
119
+ return [
120
+ datasets.SplitGenerator(
121
+ name=datasets.Split.TRAIN,
122
+ gen_kwargs={
123
+ "images": dl_manager.iter_archive(train_imgs_paths),
124
+ "masks": dl_manager.iter_archive(train_masks_paths),
125
+ },
126
+ ),
127
+ datasets.SplitGenerator(
128
+ name=datasets.Split.VALIDATION,
129
+ gen_kwargs={
130
+ "images": dl_manager.iter_archive(val_imgs_paths),
131
+ "masks": dl_manager.iter_archive(val_masks_paths),
132
+ },
133
+ ),
134
+ ]
135
+
136
+ def _generate_examples(
137
+ self,
138
+ images: Generator[Tuple[str,IO], None, None],
139
+ masks: Generator[Tuple[str,IO], None, None],
140
+ ):
141
+ for i, (img_info, mask_info) in enumerate(zip(images, masks)):
142
+ img_file_path, img_file_obj = img_info
143
+ mask_file_path, mask_file_obj = mask_info
144
+ yield i, {
145
+ "image": {"path": img_file_path, "bytes": img_file_obj.read()},
146
+ "mask": {"path": mask_file_path, "bytes": mask_file_obj.read()},
147
+ }