1aurent commited on
Commit
38f4bf8
1 Parent(s): 0c7f822

Create Human-Embryo-Dataset.py

Browse files
Files changed (1) hide show
  1. Human-Embryo-Dataset.py +206 -0
Human-Embryo-Dataset.py ADDED
@@ -0,0 +1,206 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import csv
2
+ import datasets
3
+ import pandas as pd
4
+ from pathlib import Path
5
+ from PIL import ImageFile
6
+
7
+ ImageFile.LOAD_TRUNCATED_IMAGES = True
8
+
9
+ _URLS = {
10
+ "F-45": "https://zenodo.org/records/7912264/files/embryo_dataset_F-45.tar.gz",
11
+ "F-30": "https://zenodo.org/records/7912264/files/embryo_dataset_F-30.tar.gz",
12
+ "F-15": "https://zenodo.org/records/7912264/files/embryo_dataset_F-15.tar.gz",
13
+ "F0": "https://zenodo.org/records/7912264/files/embryo_dataset.tar.gz",
14
+ "F+15": "https://zenodo.org/records/7912264/files/embryo_dataset_F15.tar.gz",
15
+ "F+30": "https://zenodo.org/records/7912264/files/embryo_dataset_F30.tar.gz",
16
+ "F+45": "https://zenodo.org/records/7912264/files/embryo_dataset_F45.tar.gz",
17
+ "grades": "https://zenodo.org/records/7912264/files/embryo_dataset_grades.csv",
18
+ "annotations": "https://zenodo.org/records/7912264/files/embryo_dataset_annotations.tar.gz",
19
+ "time_elapsed": "https://zenodo.org/records/7912264/files/embryo_dataset_time_elapsed.tar.gz",
20
+ }
21
+
22
+ _EVENT_NAMES = [
23
+ "tPB2", "tPNa", "tPNf", "t2", "t3", "t4", "t5", "t6", "t7", "t8", "t9+", "tM", "tSB", "tB", "tEB", "tHB",
24
+ ]
25
+
26
+ _GRADES = ["A", "B", "C", "NA"]
27
+
28
+ _DESCRIPTION = """
29
+ This dataset is composed of 704 videos, each recorded at 7 focal planes, accompanied by the annotations of 16 cellular events.
30
+ """
31
+
32
+ _VERSION = datasets.Version("0.3.0")
33
+
34
+ _HOMEPAGE = "https://zenodo.org/record/7912264"
35
+
36
+ _LICENSE = "CC BY-NC-SA 4.0"
37
+
38
+ class HumanEmbryoTimelapse(datasets.GeneratorBasedBuilder):
39
+
40
+ def _info(self):
41
+ return datasets.DatasetInfo(
42
+ description=_DESCRIPTION,
43
+ version=_VERSION,
44
+ homepage=_HOMEPAGE,
45
+ license=_LICENSE,
46
+ features=datasets.Features(
47
+ {
48
+ "name": datasets.Value("string"),
49
+ "F-45": datasets.Sequence(datasets.Image()),
50
+ "F-30": datasets.Sequence(datasets.Image()),
51
+ "F-15": datasets.Sequence(datasets.Image()),
52
+ "F0": datasets.Sequence(datasets.Image()),
53
+ "F+45": datasets.Sequence(datasets.Image()),
54
+ "F+30": datasets.Sequence(datasets.Image()),
55
+ "F+15": datasets.Sequence(datasets.Image()),
56
+ "events": datasets.Sequence(
57
+ {
58
+ "name": datasets.ClassLabel(names=_EVENT_NAMES),
59
+ "frame_index_start": datasets.Value("uint16"),
60
+ "frame_index_stop": datasets.Value("uint16"),
61
+ },
62
+ ),
63
+ "timeline": {
64
+ "frame_index": datasets.Sequence(datasets.Value("uint16")),
65
+ "time": datasets.Sequence(datasets.Value("float32")),
66
+ },
67
+ "grades": {
68
+ "TE": datasets.ClassLabel(names=_GRADES),
69
+ "ICM": datasets.ClassLabel(names=_GRADES),
70
+ }
71
+ }
72
+ ),
73
+ )
74
+
75
+ def _split_generators(self, dl_manager):
76
+ """Generate splits."""
77
+
78
+ # download and extract all files
79
+ directories = {
80
+ name: Path(dl_manager.download_and_extract(url))
81
+ for name, url in _URLS.items()
82
+ }
83
+
84
+ # get all subfolders of embryo_names_dir
85
+ embryo_names_dir = directories["F0"] / "embryo_dataset"
86
+ embryo_names = [x.name for x in embryo_names_dir.iterdir() if x.is_dir()]
87
+
88
+ return [
89
+ datasets.SplitGenerator(
90
+ name=datasets.Split.TRAIN,
91
+ gen_kwargs={
92
+ "embryo_names": embryo_names,
93
+ "directories": directories,
94
+ },
95
+ )
96
+ ]
97
+
98
+ def _generate_examples(self, embryo_names, directories):
99
+ """Generate images and labels for splits."""
100
+
101
+ # get grades for each embryo (name, TE, ICM)
102
+ pd_grades = pd.read_csv(directories["grades"], delimiter=',')
103
+ grades = {
104
+ row["video_name"]: {
105
+ "TE": row["TE"],
106
+ "ICM": row["ICM"],
107
+ }
108
+ for _, row in pd_grades.iterrows()
109
+ }
110
+
111
+ for index, embryo_name in enumerate(embryo_names):
112
+
113
+ # get events of the embryo (name, frame_index_start, frame_index_stop)
114
+ pd_events = pd.read_csv(directories["annotations"] / "embryo_dataset_annotations" / f"{embryo_name}_phases.csv", header=None)
115
+ events = [
116
+ {
117
+ "name": row[0],
118
+ "frame_index_start": row[1],
119
+ "frame_index_stop": row[2],
120
+ }
121
+ for _, row in pd_events.iterrows()
122
+ ]
123
+
124
+ # get frame index and time
125
+ pd_time = pd.read_csv(directories["time_elapsed"] / "embryo_dataset_time_elapsed" / f"{embryo_name}_timeElapsed.csv")
126
+ timeline = {
127
+ "frame_index": pd_time["frame_index"].tolist(),
128
+ "time": pd_time["time"].tolist(),
129
+ }
130
+
131
+ # get images of the embryo, with focal plane -45
132
+ F_m45 = list(map(
133
+ lambda x: str(x),
134
+ sorted(
135
+ (directories["F-45"] / "embryo_dataset_F-45" / embryo_name).glob("*.jpeg"),
136
+ key=lambda x: int(x.stem.split("RUN")[-1]),
137
+ ),
138
+ ))
139
+
140
+ # get images of the embryo, with focal plane -30
141
+ F_m30 = list(map(
142
+ lambda x: str(x),
143
+ sorted(
144
+ (directories["F-30"] / "embryo_dataset_F-30" / embryo_name).glob("*.jpeg"),
145
+ key=lambda x: int(x.stem.split("RUN")[-1]),
146
+ ),
147
+ ))
148
+
149
+ # get images of the embryo, with focal plane -15
150
+ F_m15 = list(map(
151
+ lambda x: str(x),
152
+ sorted(
153
+ (directories["F-15"] / "embryo_dataset_F-15" / embryo_name).glob("*.jpeg"),
154
+ key=lambda x: int(x.stem.split("RUN")[-1]),
155
+ ),
156
+ ))
157
+
158
+ # get images of the embryo, with focal plane 0
159
+ F_zero = list(map(
160
+ lambda x: str(x),
161
+ sorted(
162
+ (directories["F0"] / "embryo_dataset" / embryo_name).glob("*.jpeg"),
163
+ key=lambda x: int(x.stem.split("RUN")[-1]),
164
+ ),
165
+ ))
166
+
167
+ # get images of the embryo, with focal plane +15
168
+ F_p15 = list(map(
169
+ lambda x: str(x),
170
+ sorted(
171
+ (directories["F+15"] / "embryo_dataset_F15" / embryo_name).glob("*.jpeg"),
172
+ key=lambda x: int(x.stem.split("RUN")[-1]),
173
+ ),
174
+ ))
175
+
176
+ # get images of the embryo, with focal plane +30
177
+ F_p30 = list(map(
178
+ lambda x: str(x),
179
+ sorted(
180
+ (directories["F+30"] / "embryo_dataset_F30" / embryo_name).glob("*.jpeg"),
181
+ key=lambda x: int(x.stem.split("RUN")[-1]),
182
+ ),
183
+ ))
184
+
185
+ # get images of the embryo, with focal plane +45
186
+ F_p45 = list(map(
187
+ lambda x: str(x),
188
+ sorted(
189
+ (directories["F+45"] / "embryo_dataset_F45" / embryo_name).glob("*.jpeg"),
190
+ key=lambda x: int(x.stem.split("RUN")[-1]),
191
+ ),
192
+ ))
193
+
194
+ yield index, {
195
+ "name": embryo_name,
196
+ "F-45": F_m45,
197
+ "F-30": F_m30,
198
+ "F-15": F_m15,
199
+ "F0": F_zero,
200
+ "F+15": F_p15,
201
+ "F+30": F_p30,
202
+ "F+45": F_p45,
203
+ "events": events,
204
+ "grades": grades[embryo_name],
205
+ "timeline": timeline,
206
+ }