HMDB51 is a large human action recognition dataset consisting of 6,766 video clips (.avi) distributed across 51 action categories (e.g., jumping, drinking, kicking, etc). It is commonly used as a benchmark for video classification and action recognition tasks.
This dataset was originally collected and published by the Serre Lab at Brown University.
🔗 Serre Lab – Brown University
6,766 video clips
51 action… See the full description on the dataset page:
https://huggingface.co/datasets/hufeng7/hmdb51.
