Update README.md
Browse files
README.md
CHANGED
@@ -13,12 +13,8 @@ language:
|
|
13 |
FETV consist of a diverse set of text prompts, categorized based on three orthogonal aspects: major content, attribute control, and prompt complexity.
|
14 |
![](./Figures/categorization.png)
|
15 |
|
16 |
-
##
|
17 |
-
|
18 |
-
![](./Figures/content_attribute_statistics.png)
|
19 |
-
![](./Figures/complexity_statistics.png)
|
20 |
-
|
21 |
-
## Data Format
|
22 |
All FETV data are all available in the file `fetv_data.json`. Each line is a data instance, which is formatted as:
|
23 |
```
|
24 |
{
|
@@ -37,8 +33,14 @@ All FETV data are all available in the file `fetv_data.json`. Each line is a dat
|
|
37 |
"video_url": "https://ak.picdn.net/shutterstock/videos/1006807024/preview/stock-footage-a-mountain-stream.mp4"
|
38 |
}
|
39 |
```
|
|
|
40 |
* "video_id" is the video identifier in the original dataset where the prompt comes from.
|
41 |
* "prompt" is the text prompt for text-to-video generation.
|
42 |
* "major content", "attribute control" and "prompt complexity" are the three orthogonal aspects for categorization.
|
43 |
* "source" denotes the original dataset where the prompt comes from, which can be "WebVid", "MSRVTT" or "ours".
|
44 |
-
* "video_url" is the url link of the reference video.
|
|
|
|
|
|
|
|
|
|
|
|
13 |
FETV consist of a diverse set of text prompts, categorized based on three orthogonal aspects: major content, attribute control, and prompt complexity.
|
14 |
![](./Figures/categorization.png)
|
15 |
|
16 |
+
## Dataset Structure
|
17 |
+
### Data Instances
|
|
|
|
|
|
|
|
|
18 |
All FETV data are all available in the file `fetv_data.json`. Each line is a data instance, which is formatted as:
|
19 |
```
|
20 |
{
|
|
|
33 |
"video_url": "https://ak.picdn.net/shutterstock/videos/1006807024/preview/stock-footage-a-mountain-stream.mp4"
|
34 |
}
|
35 |
```
|
36 |
+
### Data Fields
|
37 |
* "video_id" is the video identifier in the original dataset where the prompt comes from.
|
38 |
* "prompt" is the text prompt for text-to-video generation.
|
39 |
* "major content", "attribute control" and "prompt complexity" are the three orthogonal aspects for categorization.
|
40 |
* "source" denotes the original dataset where the prompt comes from, which can be "WebVid", "MSRVTT" or "ours".
|
41 |
+
* "video_url" is the url link of the reference video.
|
42 |
+
|
43 |
+
### Dataset Statistics
|
44 |
+
FETV contains 619 text prompts. The data distributions over different categories are as follows
|
45 |
+
![](./Figures/content_attribute_statistics.png)
|
46 |
+
![](./Figures/complexity_statistics.png)
|