Upload README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,48 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
## Video Augmented Texts Data
|
| 2 |
+
|
| 3 |
+
### VATEX
|
| 4 |
+
|
| 5 |
+
Each video contains 10 captions. In `vatex.zip`, there are:
|
| 6 |
+
|
| 7 |
+
* `test/`: a folder containing all available videos
|
| 8 |
+
* `vatex_public_test_english_v1.1.json`: JSON file containing all captions
|
| 9 |
+
|
| 10 |
+
Example data loading:
|
| 11 |
+
|
| 12 |
+
```py
|
| 13 |
+
import os
|
| 14 |
+
import json
|
| 15 |
+
|
| 16 |
+
path = 'vatex_public_test_english_v1.1.json'
|
| 17 |
+
d = json.load(open(path, 'r'))
|
| 18 |
+
|
| 19 |
+
captions = {v['videoID']: v['enCap'] for v in d}
|
| 20 |
+
|
| 21 |
+
for vname in captions:
|
| 22 |
+
video_path = os.path.join('test', vname+'.mp4') # path to the video
|
| 23 |
+
captions = captions[vname] # a list of 10 str
|
| 24 |
+
```
|
| 25 |
+
|
| 26 |
+
### MSR-VTT
|
| 27 |
+
|
| 28 |
+
Each video contains 1 caption. There are two files for MSR-VTT:
|
| 29 |
+
|
| 30 |
+
* `MSRVTT.zip`: contains all videos
|
| 31 |
+
* `MSRVTT_JSFUSION_test.csv`: contains all captions
|
| 32 |
+
|
| 33 |
+
Example data loading:
|
| 34 |
+
|
| 35 |
+
```py
|
| 36 |
+
import os
|
| 37 |
+
import pandas as pd
|
| 38 |
+
|
| 39 |
+
path = 'MSRVTT_JSFUSION_test.csv'
|
| 40 |
+
df = pd.read_csv(path)
|
| 41 |
+
|
| 42 |
+
vid_id_list = df['video_id'].tolist()
|
| 43 |
+
caption_list = df['sentence'].tolist()
|
| 44 |
+
|
| 45 |
+
for vid_id, caption in zip(vid_id_list, caption_list):
|
| 46 |
+
video_path = os.path.join('MSRVTT', 'videos', 'all', vid_id+'.mp4')
|
| 47 |
+
captions = [caption] # a list of 1 str
|
| 48 |
+
```
|