1 dataset found

Tags: atomic visual actions

Filter Results
  • AVA v2.2

    The AVA v2.2 dataset for spatiotemporal action localization contains the bounding box annotations and the corresponding action labels on keyframes.