|
# TAO annotations |
|
|
|
Files: |
|
- `{train,validation}.json`: Contains annotations for LVIS categories in TAO |
|
train and validation. |
|
- `{train,validation}_with_freeform.json`: Contains annotations for all |
|
categories in TAO train and validation. |
|
- `test_without_annotations.json`: JSON for test videos. The 'images' and |
|
'videos' fields contain the images and videos that will be evaluated on the |
|
test set. |
|
- `test_categories.json`: Contains a list of categories which will be evaluated |
|
on the TAO test set. |
|
- `video-lists/`: Contains lists of videos and their source datasets contained |
|
in TAO. Please do not train on any videos in these lists! See |
|
`video-lists/README.md` for details. |
|
|
