Action Parsing

Action Parsing is a sub-task in the field of computer vision that aims to annotate actions in every frame of a video or in static images. This task achieves fine-grained analysis of dynamic scenes by recognizing and classifying actions at each time point, providing critical support for applications such as behavior understanding, pose estimation, and anomaly detection.