faizdevx/ActionSnap

ActionSnap is an activity recognition model implemented based on the Transformer neural network model architecture using 2D human poses. ActionSnap understands and recognizes human actions by analyzing movements and patterns obtained from 2D human pose landmarks that have been obtained using the MediaPipe Pose module.

语言：Jupyter Notebook · GitHub 星标：16 · Fork：0

在 GitHub 上查看原仓库