3D Hand Pose Estimation Based on Deep Learning and Ensemble Learning

A 3D hand pose estimation method is proposed based on ensemble learning for fixed hand postures. First, a data set with 3D attitude angle tags for fixed hand posture is constructed; Then, a single convolutional neural network is used to fit the hand image and the corresponding 3D attitude angle label; Finally, two different ensemble learning strategies are proposed to improve the performance of a single CNN hand pose estimation model. The experimental results verify the effectiveness and practicability of this hand pose estimation method.