> For the complete documentation index, see [llms.txt](https://docs.roboflow.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.roboflow.com/reference/changelog/changelog/september-2026/action-recognition-upload-label-and-train-on-video.md).

# Action recognition: Upload, label, and train on video

Action recognition is now available as a project type in Roboflow. Upload video clips, drag across the timeline to mark the time range where each action happens, and train a model that reads several frames at once and scores the action across them. The first supported model is NVIDIA's Cosmos 3 Edge, with more on the way.

[Learn more about annotating video segments](https://docs.roboflow.com/datasets/annotate/annotate/annotate-video-segments)