Dataset Viewer
Auto-converted to Parquet
Search is not available for this dataset
video
video
label
class label
51 classes
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
0brush_hair
End of preview. Expand in Data Studio

HMDB51 Dataset

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.

πŸ“¦ Dataset Source

This dataset was originally collected and published by the Serre Lab at Brown University.

πŸ”— Serre Lab – Brown University

πŸ“ Contents

  • 6,766 video clips
  • 51 action categories
  • Includes training/testing splits for benchmarking

πŸ“š Citation

If you use this dataset, please cite:

Kuehne, H., Jhuang, H., Garrote, E., Poggio, T., & Serre, T. (2011). HMDB: A large video database for human motion recognition. In Proceedings of the IEEE International Conference on Computer Vision (ICCV).

Downloads last month
42
Papers with Code