Lightweight Gesture Pose Estimation Based on CPM Algorithm

With the experiential enhancement of artificial intelligence products, gesture estimation, as a classic computer vision task, has wide range application scenarios. Aiming at current network model that needs to be lightweight in mobile smart this paper designs pose estimation based on CPM (Convolutional Pose Machine) multi-stage human network. A comparative experiment RHD open-source data set was conducted compare and analyze while ensuring accuracy effectively reducing amount parameters, which provides basis for development real-time terminal estimation.