Okutama Action Dataset is a video dataset for concurrent human action detection. It contains 43 minutes of fully annotated sequences and 12 action categories. The dataset can be used for a variety of tasks: pedestrian detection, spatiotemporal action detection, and multi-person tracking.

The Okutama Action Dataset was released by the National Institute of Informatics in Tokyo in 2017. The related paper is "Okutama-Action: An Aerial View Video Dataset for Concurrent Human Action Detection".