root
add annotations
61b5098
raw
history blame
597 Bytes
Countix Dataset
Data is organized as follows:
1) countix_train.csv - Train split
2) countix_val.csv - Validation split
3) countix_test.csv - Test split
Each csv file has the following columns:
1) video_id - YouTube video ID.
2) class - Kinetics action class.
3) kinetics_start - Time in seconds when Kinetics clip starts.
4) kinetics_end - Time in seconds when Kinetics clip ends.
5) repetition_start - Time in seconds when repetitions begin in the video.
6) repetition_end - Time in seconds when repetitions end in the video.
7) count - The number of times the action was repeated in the video.