Estimating 2D Multi-hand Poses from Single Depth Images

Le Duan, Minmin Shen, Song Cui, Zhexiao Guo, Oliver Deussen

Abstract: We present a novel framework based on Pictorial Structure (PS) models to estimate 2D multi-hand poses from depth images. Most existing single-hand pose estimation algorithms are either subject to strong assumptions or depend on a weak detector to detect the human hand. We utilize Mask R-CNN to avoid both aforementioned constraints. The proposed framework allows detection of multi-hand instances and localization of hand joints simultaneously. Our experiments show that our method is superior to existing methods.