holylovenia commited on
Commit
386b31f
1 Parent(s): 766b71c

Upload fsl_105.py with huggingface_hub

Browse files
Files changed (1) hide show
  1. fsl_105.py +194 -0
fsl_105.py ADDED
@@ -0,0 +1,194 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ from pathlib import Path, PureWindowsPath
3
+ from typing import Dict, List, Tuple
4
+
5
+ try:
6
+ import cv2
7
+ except:
8
+ print("Install the `cv2` package to use.")
9
+ import datasets
10
+ import pandas as pd
11
+
12
+ from seacrowd.utils import schemas
13
+ from seacrowd.utils.configs import SEACrowdConfig
14
+ from seacrowd.utils.constants import Licenses, Tasks
15
+
16
+ _CITATION = """\
17
+ @article{tupal4476867fsl105,
18
+ title={FSL105: The Video Filipino Sign Language Sign Database of Introductory 105 FSL Signs},
19
+ author={Tupal, Isaiah Jassen Lizaso and Melvin, Cabatuan K},
20
+ journal={Available at SSRN 4476867}
21
+ }
22
+ """
23
+
24
+ _DATASETNAME = "fsl_105"
25
+
26
+ _DESCRIPTION = """\
27
+ FSL-105 is a video dataset for 105 different Filipino Sign Language (FSL) signs.
28
+ Each sign is categorized into one of 10 categories and is each represented by approximately 20 four-second video samples.
29
+ Signs were performed by adult deaf FSL signers on a blank blue background and reviewed by an FSL expert.
30
+ """
31
+
32
+ _HOMEPAGE = "https://data.mendeley.com/datasets/48y2y99mb9/2"
33
+
34
+ _LICENSE = Licenses.CC_BY_4_0.value
35
+
36
+ _LOCAL = False
37
+
38
+ _URLS = {
39
+ "clips": "https://prod-dcd-datasets-public-files-eu-west-1.s3.eu-west-1.amazonaws.com/de95a3c3-02f4-4a3f-9a9e-ce2371160275",
40
+ "train": "https://prod-dcd-datasets-public-files-eu-west-1.s3.eu-west-1.amazonaws.com/09c71779-3a2a-4c98-8d9b-0ef74f54d92a",
41
+ "test": "https://prod-dcd-datasets-public-files-eu-west-1.s3.eu-west-1.amazonaws.com/39af8117-6b44-47b9-a551-0bdc40837295",
42
+ }
43
+
44
+ _LANGUAGES = ["psp"]
45
+
46
+ _SUPPORTED_TASKS = [Tasks.VIDEO_TO_TEXT_RETRIEVAL, Tasks.VIDEO_CAPTIONING]
47
+
48
+ _SOURCE_VERSION = "1.0.0"
49
+
50
+ _SEACROWD_VERSION = "2024.06.20"
51
+
52
+
53
+ class FSL105Dataset(datasets.GeneratorBasedBuilder):
54
+ """
55
+ FSL-105 is a video dataset for 105 different Filipino Sign Language (FSL) signs.
56
+ Each sign is categorized into one of 10 categories and is each represented by approximately 20 four-second video samples.
57
+ Signs were performed by adult deaf FSL signers on a blank blue background and reviewed by an FSL expert.
58
+ """
59
+
60
+ SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
61
+ SEACROWD_VERSION = datasets.Version(_SEACROWD_VERSION)
62
+
63
+ BUILDER_CONFIGS = [
64
+ SEACrowdConfig(
65
+ name=f"{_DATASETNAME}_source",
66
+ version=SOURCE_VERSION,
67
+ description=f"{_DATASETNAME} source schema",
68
+ schema="source",
69
+ subset_id=f"{_DATASETNAME}",
70
+ ),
71
+ SEACrowdConfig(
72
+ name=f"{_DATASETNAME}_seacrowd_vidtext",
73
+ version=SEACROWD_VERSION,
74
+ description=f"{_DATASETNAME} SEACrowd schema",
75
+ schema="seacrowd_vidtext",
76
+ subset_id=f"{_DATASETNAME}",
77
+ ),
78
+ ]
79
+
80
+ DEFAULT_CONFIG_NAME = f"{_DATASETNAME}_source"
81
+
82
+ category = [
83
+ "CALENDAR",
84
+ "COLOR",
85
+ "DAYS",
86
+ "DRINK",
87
+ "FAMILY",
88
+ "FOOD",
89
+ "GREETING",
90
+ "NUMBER",
91
+ "RELATIONSHIPS",
92
+ "SURVIVAL",
93
+ ]
94
+
95
+ def _info(self) -> datasets.DatasetInfo:
96
+ if self.config.schema == "source":
97
+ features = datasets.Features(
98
+ {
99
+ "id": datasets.Value("string"),
100
+ "video_path": datasets.Value("string"),
101
+ "text": datasets.Value("string"),
102
+ "labels": datasets.ClassLabel(names=self.category),
103
+ "metadata": {
104
+ "resolution": {
105
+ "width": datasets.Value("int64"),
106
+ "height": datasets.Value("int64"),
107
+ },
108
+ "duration": datasets.Value("float32"),
109
+ "fps": datasets.Value("float32"),
110
+ },
111
+ }
112
+ )
113
+
114
+ elif self.config.schema == "seacrowd_vidtext":
115
+ features = schemas.video_features
116
+
117
+ return datasets.DatasetInfo(
118
+ description=_DESCRIPTION,
119
+ features=features,
120
+ homepage=_HOMEPAGE,
121
+ license=_LICENSE,
122
+ citation=_CITATION,
123
+ )
124
+
125
+ def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
126
+ """Returns SplitGenerators."""
127
+
128
+ clips = dl_manager.download_and_extract(_URLS["clips"])
129
+ train = dl_manager.download_and_extract(_URLS["train"])
130
+ test = dl_manager.download_and_extract(_URLS["test"])
131
+
132
+ train_df = pd.read_csv(train)
133
+ test_df = pd.read_csv(test)
134
+
135
+ return [
136
+ datasets.SplitGenerator(
137
+ name=datasets.Split.TRAIN,
138
+ gen_kwargs={
139
+ "filepath": {
140
+ "clips": clips,
141
+ "data": train_df,
142
+ },
143
+ "split": "train",
144
+ },
145
+ ),
146
+ datasets.SplitGenerator(
147
+ name=datasets.Split.TEST,
148
+ gen_kwargs={
149
+ "filepath": {"clips": clips, "data": test_df},
150
+ "split": "test",
151
+ },
152
+ ),
153
+ ]
154
+
155
+ def _generate_examples(self, filepath: Path, split: str) -> Tuple[int, Dict]:
156
+ """Yields examples as (key, example) tuples."""
157
+
158
+ for key, example in filepath["data"].iterrows():
159
+ video = cv2.VideoCapture(os.path.join(filepath["clips"], PureWindowsPath(example["vid_path"]).as_posix()))
160
+ fps = video.get(cv2.CAP_PROP_FPS)
161
+ frame_count = video.get(cv2.CAP_PROP_FRAME_COUNT)
162
+ duration = frame_count / fps
163
+ vid_width = video.get(cv2.CAP_PROP_FRAME_WIDTH)
164
+ vid_height = video.get(cv2.CAP_PROP_FRAME_HEIGHT)
165
+
166
+ if self.config.schema == "source":
167
+ yield key, {
168
+ "id": str(key),
169
+ "video_path": os.path.join(filepath["clips"], example["vid_path"]),
170
+ "text": example["label"],
171
+ "labels": example["category"],
172
+ "metadata": {
173
+ "resolution": {
174
+ "width": vid_width,
175
+ "height": vid_height,
176
+ },
177
+ "duration": duration,
178
+ "fps": fps,
179
+ },
180
+ }
181
+ elif self.config.schema == "seacrowd_vidtext":
182
+ yield key, {
183
+ "id": str(key),
184
+ "video_path": os.path.join(filepath["clips"], example["vid_path"]),
185
+ "text": example["label"],
186
+ "metadata": {
187
+ "resolution": {
188
+ "width": vid_width,
189
+ "height": vid_height,
190
+ },
191
+ "duration": duration,
192
+ "fps": fps,
193
+ },
194
+ }