Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,55 @@
|
|
1 |
-
---
|
2 |
-
license: mit
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: mit
|
3 |
+
task_categories:
|
4 |
+
- image-feature-extraction
|
5 |
+
- video-classification
|
6 |
+
language:
|
7 |
+
- en
|
8 |
+
---
|
9 |
+
|
10 |
+
# Generating_Video_Data
|
11 |
+
|
12 |
+
```
|
13 |
+
Hanzhe Liang
|
14 |
+
modified: 23 May
|
15 |
+
```
|
16 |
+
|
17 |
+
😊 This library contains some common data for training video generation models. This is a library of summary datasets, with thanks to the original authors for their contributions. Regardless of which data you need, the following steps:
|
18 |
+
```bash
|
19 |
+
mkdir data
|
20 |
+
cd data
|
21 |
+
```
|
22 |
+
Next, go to my cloud database link and download the data you want into the folder```data/```.
|
23 |
+
Once you've downloaded the data you need, we've found preprocessing steps from its official implementation, which you can follow to process the data.
|
24 |
+
For memory requirements, we provide the files as single zip files, you need to extract the file named data_name.rar with the unzip command: ```unrar x data_name.rar```
|
25 |
+
|
26 |
+
----
|
27 |
+
|
28 |
+
### Something Something V2
|
29 |
+
Please check the data path.
|
30 |
+
```bash
|
31 |
+
cd somethingv2
|
32 |
+
python extract_frames.py
|
33 |
+
python process_somethingv2.py
|
34 |
+
```
|
35 |
+
You can check data list ```somethingv2/*.txt```
|
36 |
+
|
37 |
+
### Human3.6M
|
38 |
+
I provided two datasets, both processed and unprocessed.
|
39 |
+
|
40 |
+
Unprocessed
|
41 |
+
```bash
|
42 |
+
cd human36m/human36m_unprocessed
|
43 |
+
python build_clip_dataset.py
|
44 |
+
python make_list.py
|
45 |
+
```
|
46 |
+
You can check data list ```human36m/human36m_unprocessed/*.txt```
|
47 |
+
|
48 |
+
or you can directly check the processed data in ```human36m/human36m_processed```
|
49 |
+
|
50 |
+
### YouTubeDriving
|
51 |
+
```bash
|
52 |
+
cd youtubedriving
|
53 |
+
python make_list.py
|
54 |
+
```
|
55 |
+
You can check data list ```youtubedriving/*.txt```
|