Hugging Face's logo

Datasets:
lixinhao
/
VideoEval

Tasks:
Feature Extraction
Modalities:
Text
Formats:
text
Size:
10K - 100K
ArXiv:
Libraries:
Datasets
License:
Dataset card Data Studio Files
xet
Community
1
VideoEval
File size: 4,334 Bytes
8b86d70
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
 
---
license: apache-2.0
task_categories:
- feature-extraction
size_categories:
- 10K<n<100K
---

# Dataset Card for VideoEval

## VidTAB

### Action Recognition in Dark

You could download all videos from ARID at https://opendatalab.com/OpenDataLab/Action_Recognition_in_the_Dark.

You just need to use the mp4 video in the video folder and then use the [annotations](https://github.com/leexinhao/VideoEval/tree/main/VidTAB/annotations/AR_in_Dark) we provided.

### Action Recognition in Long Video

You could download all videos from Breakfast at https://serre-lab.clps.brown.edu/resource/breakfast-actions-dataset/.

You just need to use the mp4 video in the video folder and then use the [annotations](https://github.com/leexinhao/VideoEval/tree/main/VidTAB/annotations/AR_in_Long) we provided.

### Medical Surgery
You could download all videos from SurgicalActions160 at http://ftp.itec.aau.at/datasets/SurgicalActions160/index.html. 

You just need to use the mp4 video in the video folder and then use the [annotations](https://github.com/leexinhao/VideoEval/tree/main/VidTAB/annotations/Medical_Surgery)  we provided.

### Animal Behavior

You could download all videos from Animal Kingdom at https://forms.office.com/pages/responsepage.aspx?id=drd2NJDpck-5UGJImDFiPVRYpnTEMixKqPJ1FxwK6VZUQkNTSkRISTNORUI2TDBWMUpZTlQ5WUlaSyQlQCN0PWcu.

You just need to use the mp4 video in the video folder and then use the [annotations](https://github.com/leexinhao/VideoEval/tree/main/VidTAB/annotations/Animal_Behavior)  we provided.

### Harmful Content

You could download all videos from MOB at https://drive.google.com/file/d/1Zjib-WaF5hk3wVrj5eW6ewdpMFcn45Wo/view.

Merge folders benign and malicious and then use the  [annotations](https://github.com/leexinhao/VideoEval/tree/main/VidTAB/annotations/Quality_Access) we provided.

### Fake Face

You could download all videos from FaceForensics++ at https://docs.google.com/forms/d/e/1FAIpQLSdRRR3L5zAv6tQ_CKxmK4W96tAab_pfBu2EKAgQbeDVhmXagg/viewform?pli=1.

Then

```bash
cd yourpath/FaceForensics++
mkdir videos
mv faceforensics_videos/original_sequences/youtube/c23 videos/pos
mkdir videos/neg
python get_negs_samples.py
```

`get_negs_samples.py` is 

```python
import os
import shutil

video_list = os.listdir('videos/pos')
assert len(video_list) == 1000, len(video_list) 

for i in range(0, 1000):
    for method in ["Deepfakes", "Face2Face", "FaceShifter", "FaceSwap", "NeuralTextures"]:
    	shutil.copy(f"faceforensics_videos/manipulated_sequences/{method}/c23/videos/{video_list[i]}", f"videos/neg/{video_list[i][:-4]}-{method}.mp4")
```

And then use the  [annotations](https://github.com/leexinhao/VideoEval/tree/main/VidTAB/annotations/Fake_Face) we provided.


### Emotion Analysis

You could download all videos from CAER at https://drive.google.com/file/d/1JsdbBkulkIOqrchyDnML2GEmuwi6E_d2/view

You just need to use the mp4 video in the video folder and then use the [annotations](https://github.com/leexinhao/VideoEval/tree/main/VidTAB/annotations/Emotion_Analysis) we provided.

### Quality Access

You could download all videos from DOVER at  https://huggingface.co/datasets/teowu/DIVIDE-MaxWell/resolve/main/videos.zip.

You just need to use the mp4 video in the video folder and then use the [annotations](https://github.com/leexinhao/VideoEval/tree/main/VidTAB/annotations/Quality_Access) we provided.

## VidEB

### FIVR-5K

* Install [yt-dlp](https://github.com/yt-dlp/yt-dlp) (make sure it is up-to-date)
* Run the following command to download videos:

```bash
python VidEB/annotations/FIVR-5K/download_dataset.py \
						   --video_dir VIDEO_DIR \
						   --dataset_ids VidEB/annotations/FIVR-5K/used_videos.txt \
						   --cores NUMBER_OF_CODES \
						   --resolution RESOLUTION
```

### DVSC23

For queries,

```bash
wget -i VidEB/annotations/DVSC23/vsc_queries.txt --cut-dirs 2 -x -nH
```

For database,

```bash
wget -i VidEB/annotations/DVSC23/vsc_database.txt --cut-dirs 2 -x -nH
```

## Citation

**BibTeX:**

```
@article{li2024videoeval,
  title={Videoeval: Comprehensive benchmark suite for low-cost evaluation of video foundation model},
  author={Li, Xinhao and Huang, Zhenpeng and Wang, Jing and Li, Kunchang and Wang, Limin},
  journal={arXiv preprint arXiv:2407.06491},
  year={2024}
}
```


## Dataset Card Contact

xinhaoli00@outlook.com