File size: 7,788 Bytes
0913a62
 
c40436c
 
a83f92d
c40436c
 
e150a9f
c40436c
e0f2707
e150a9f
 
e0f2707
 
 
a83f92d
e150a9f
 
 
 
27cc8ee
 
 
 
 
 
0b30aac
 
 
41e4dc9
283d943
2d3204d
c96da1e
 
 
 
6ccce9b
283d943
e9ee83a
41e4dc9
7ce248b
 
b553e2e
48e90ae
 
 
a1afccc
 
 
48e90ae
 
 
 
 
 
 
 
283d943
 
 
 
 
 
 
 
 
 
 
 
7917ad0
48e90ae
 
 
7917ad0
0f3e420
 
e72c0ae
3182d09
 
 
 
 
 
 
 
 
 
0f3e420
9faa859
3182d09
5547469
3182d09
 
 
0f3e420
 
 
53b1cda
 
02145a6
 
 
 
8e176e5
02145a6
 
 
 
 
 
 
 
 
 
53b1cda
 
c96da1e
53b1cda
02145a6
5d44dc2
 
 
 
 
 
 
 
 
 
 
 
 
8e176e5
53b1cda
 
 
e3532f3
d1f7e14
a1afccc
0f3e420
47cb437
 
 
0f3e420
ebb4790
69ee0e0
 
 
 
ef467ab
 
320794e
0f3e420
a63d9a1
 
 
47cb437
 
 
 
 
 
 
a63d9a1
a47a2c0
 
14ae994
22a428e
14ae994
 
 
7f5e8a2
14ae994
 
b553e2e
e9ee83a
 
 
 
 
 
 
 
b553e2e
53b1cda
b553e2e
53b1cda
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
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
---
license: cc-by-nc-4.0
task_categories:
- text-to-video
- text-to-image
language:
- en
pretty_name: VidProM
size_categories:
- 1M<n<10M
source_datasets:
  - original
tags:
- prompts
- text-to-video
- text-to-image
- Pika
- VideoCraft2
- Text2Video-Zero
- ModelScope
- Video Generative Model Evaluation
- Text-to-Video Diffusion Model Development
- Text-to-Video Prompt Engineering
- Efficient Video Generation
- Fake Video Detection
- Video Copy Detection for Diffusion Models
configs:
- config_name: VidProM_unique
  data_files: VidProM_unique.csv
---


<p align="center">
  <img src="https://huggingface.co/datasets/WenhaoWang/VidProM/resolve/main/teasor.png" width="800">
</p>


# Summary
This is the dataset proposed in our paper "[**VidProM: A Million-scale Real Prompt-Gallery Dataset for Text-to-Video Diffusion Models**](https://arxiv.org/abs/2403.06098)"

VidProM is the first dataset featuring 1.67 million unique text-to-video prompts and 6.69 million videos generated from 4 different state-of-the-art diffusion models.
It inspires many exciting new research areas, such as Text-to-Video Prompt Engineering, Efficient Video Generation, Fake Video Detection, and Video Copy Detection for Diffusion Models.

# Directory
```
*DATA_PATH
    *VidProM_unique.csv
    *VidProM_semantic_unique.csv
    *VidProM_embed.hdf5
	*original_files
		*generate_1_ori.html
		*generate_2_ori.html
        ...
	*pika_videos
		*pika_videos_1.tar
		*pika_videos_2.tar
		...
    *vc2_videos
        *vc2_videos_1.tar
		*vc2_videos_2.tar
		...
    *t2vz_videos
        *t2vz_videos_1.tar
		*t2vz_videos_2.tar
		...
    *ms_videos
        *ms_videos_1.tar
		*ms_videos_2.tar
		...
    *example

```


# Download 

### Automatical
Install the [datasets](https://huggingface.co/docs/datasets/v1.15.1/installation.html) library first, by:
```
pip install datasets
```
Then it can be downloaded automatically with
```
import numpy as np
from datasets import load_dataset
dataset = load_dataset('WenhaoWang/VidProM')
```

### Manual

You can also download each file by ```wget```, for instance:
```
wget https://huggingface.co/datasets/WenhaoWang/VidProM/resolve/main/VidProM_unique.csv
```

# Explanation

``VidProM_unique.csv`` contains the UUID, prompt, time, and 6 NSFW probabilities.

It can easily be read by

```
import pandas
df = pd.read_csv("VidProM_unique.csv")
```

Below are three rows from ``VidProM_unique.csv``:
| uuid                                 | prompt                                                                                                                                                                 | time                     | toxicity | obscene | identity_attack | insult  | threat  | sexual_explicit |
|--------------------------------------|------------------------------------------------------------------------------------------------------------------------------------------------------------------------|--------------------------|----------|---------|-----------------|---------|---------|-----------------|
| 6a83eb92-faa0-572b-9e1f-67dec99b711d | Flying among clouds and stars, kitten Max discovered a world full of winged friends. Returning home, he shared his stories and everyone smiled as they imagined flying together in their dreams. | Sun Sep  3 12:27:44 2023 | 0.00129  | 0.00016 | 7e-05           | 0.00064 | 2e-05   | 2e-05           |
| 3ba1adf3-5254-59fb-a13e-57e6aa161626 | Use a clean and modern font for the text "Relate Reality 101." Add a small, stylized heart icon or a thought bubble above or beside the text to represent emotions and thoughts. Consider using a color scheme that includes warm, inviting colors like deep reds, soft blues, or soothing purples to evoke feelings of connection and intrigue. | Wed Sep 13 18:15:30 2023 | 0.00038  | 0.00013 | 8e-05           | 0.00018 | 3e-05   | 3e-05           |
| 62e5a2a0-4994-5c75-9976-2416420526f7 | zoomed out, sideview of an Grey Alien sitting at a computer desk                                                                                                       | Tue Oct 24 20:24:21 2023 | 0.01777  | 0.00029 | 0.00336         | 0.00256 | 0.00017 | 5e-05           |


``VidProM_semantic_unique.csv`` is a semantically unique version of ``VidProM_unique.csv``.

``VidProM_embed.hdf5`` is the 3072-dim embeddings of our prompts. They are embedded by text-embedding-3-large, which is the latest text embedding model of OpenAI.

It can easily be read by

```
import numpy as np
import h5py
def read_descriptors(filename):
    hh = h5py.File(filename, "r")
    descs = np.array(hh["embeddings"])
    names = np.array(hh["uuid"][:], dtype=object).astype(str).tolist()
    return names, descs

uuid, features = read_descriptors('VidProM_embed.hdf5')
```

``original_files`` are the HTML files from [official Pika Discord](https://discord.com/invite/pika) collected by [DiscordChatExporter](https://github.com/Tyrrrz/DiscordChatExporter). You can do whatever you want with it under [CC BY-NC 4.0 license](https://creativecommons.org/licenses/by-nc/4.0/deed.en).

``pika_videos``, ``vc2_videos``, ``t2vz_videos``, and ``ms_videos`` are the generated videos by 4 state-of-the-art text-to-video diffusion models. Each contains 30 tar files.

``example`` is a subfolder which contains 10,000 datapoints.


# Datapoint
<p align="center">
  <img src="https://huggingface.co/datasets/WenhaoWang/VidProM/resolve/main/datapoint.png" width="800">
</p>

# Visualization

<p align="center">
  <img src="https://huggingface.co/datasets/WenhaoWang/VidProM/resolve/main/WizMap.png" width="800">
</p>

Click the [WizMap](https://poloclub.github.io/wizmap/?dataURL=https%3A%2F%2Fhuggingface.co%2Fdatasets%2FWenhaoWang%2FVidProM%2Fresolve%2Fmain%2Fdata_gpu13.ndjson&gridURL=https%3A%2F%2Fhuggingface.co%2Fdatasets%2FWenhaoWang%2FVidProM%2Fresolve%2Fmain%2Fgrid_gpu13.json)
(and wait for 5 seconds) for an interactive visualization of our 1.67 million prompts. Above is a thumbnail.

# Comparison with DiffusionDB


<p align="center">
  <img src="https://huggingface.co/datasets/WenhaoWang/VidProM/resolve/main/compare_table.png" width="800">
</p>

<p align="center">
  <img src="https://huggingface.co/datasets/WenhaoWang/VidProM/resolve/main/compare_visual.png" width="800">
</p>

Please check our paper for a detailed comparison.

# Curators
VidProM is created by [Wenhao Wang](https://wangwenhao0716.github.io/) and Professor [Yi Yang](https://scholar.google.com/citations?user=RMSuNFwAAAAJ&hl=zh-CN) from [the ReLER Lab](https://reler.net/).

# License

The prompts and videos generated by [Pika](https://discord.com/invite/pika) in our VidProM are licensed under the [CC BY-NC 4.0 license](https://creativecommons.org/licenses/by-nc/4.0/deed.en). Additionally, similar to their original repositories, the videos from [VideoCraft2](https://github.com/AILab-CVC/VideoCrafter), [Text2Video-Zero](https://github.com/Picsart-AI-Research/Text2Video-Zero), and [ModelScope](https://huggingface.co/ali-vilab/modelscope-damo-text-to-video-synthesis) are released under the [Apache license](https://www.apache.org/licenses/LICENSE-2.0), the [CreativeML Open RAIL-M license](https://github.com/Picsart-AI-Research/Text2Video-Zero/blob/main/LICENSE), and the [CC BY-NC 4.0 license](https://creativecommons.org/licenses/by-nc/4.0/deed.en), respectively. Our code is released under the [CC BY-NC 4.0 license](https://creativecommons.org/licenses/by-nc/4.0/deed.en).


# Citation
```
@article{wang2024vidprom,
  title={VidProM: A Million-scale Real Prompt-Gallery Dataset for Text-to-Video Diffusion Models},
  author={Wang, Wenhao and Yang, Yi},
  journal={arXiv preprint arXiv:2403.06098},
  year={2024}
}
```

# Contact

If you have any questions, feel free to contact Wenhao Wang (wangwenhao0716@gmail.com).