Alkemi commited on
Commit
eb4ee4c
1 Parent(s): e38355c

Update xd-violence-small.py

Browse files
Files changed (1) hide show
  1. xd-violence-small.py +70 -0
xd-violence-small.py CHANGED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import datasets
3
+ import numpy as np
4
+
5
+ class YourDataset(datasets.GeneratorBasedBuilder):
6
+ VERSION = datasets.Version("1.0.0")
7
+
8
+ BUILDER_CONFIGS = [
9
+ datasets.BuilderConfig(
10
+ name="i3d_rgb",
11
+ version=VERSION,
12
+ description="i3d RGB dataset",
13
+ features=datasets.Features({"video": datasets.Sequence(datasets.Value("float32"))}),
14
+ ),
15
+ datasets.BuilderConfig(
16
+ name="i3d_flow",
17
+ version=VERSION,
18
+ description="i3d Flow dataset",
19
+ features=datasets.Features({"video": datasets.Sequence(datasets.Value("float32"))}),
20
+ ),
21
+ datasets.BuilderConfig(
22
+ name="vggish_train",
23
+ version=VERSION,
24
+ description="VGGish Train dataset",
25
+ features=datasets.Features({"audio": datasets.Sequence(datasets.Value("float32"))}),
26
+ ),
27
+ datasets.BuilderConfig(
28
+ name="vggish_test",
29
+ version=VERSION,
30
+ description="VGGish Test dataset",
31
+ features=datasets.Features({"audio": datasets.Sequence(datasets.Value("float32"))}),
32
+ ),
33
+ ]
34
+
35
+ def _info(self):
36
+ features = self.config.features
37
+ return datasets.DatasetInfo(
38
+ features=features,
39
+ )
40
+
41
+ def _split_generators(self, dl_manager):
42
+ data_dir = "Alkemi/xd-violence-small"
43
+
44
+ splits = []
45
+ for config in self.configs:
46
+ if config.name.startswith("i3d-features"):
47
+ split_name = config.name.split("_")[1]
48
+ splits.append(
49
+ datasets.SplitGenerator(name=split_name, gen_kwargs={"data_dir": os.path.join(data_dir, "i3d-features", split_name)})
50
+ )
51
+ elif config.name.startswith("vggish-features"):
52
+ split_name = config.name.split("_")[1]
53
+ splits.append(
54
+ datasets.SplitGenerator(name=split_name, gen_kwargs={"data_dir": os.path.join(data_dir, "vggish-features", split_name)})
55
+ )
56
+
57
+ return splits
58
+
59
+ def _generate_examples(self, data_dir):
60
+ filepaths = []
61
+
62
+ for root, dirs, files in os.walk(data_dir):
63
+ for file in files:
64
+ if file.endswith(".npy"):
65
+ filepaths.append(os.path.join(root, file))
66
+
67
+ for i, filepath in enumerate(filepaths):
68
+ data = np.load(filepath)
69
+
70
+ yield i, {"video": data} # or {"audio": data} depending on the configuration