Experimental Study of a Deep-Learning RGB-D Tracker for Virtual Remote Human Model Reconstruction

Figure 10

Visualization of the proposed point cloud compensation algorithm based on the deep-learning body tracker model for a sample case where the right wrist joint is missing in frames 61, 62, and 64 and the case where the right wrist joints have an inaccurate depth value. Unprocessed skeleton data are in green, and the estimated skeleton data are shown in multiple colors.