**

Multi-perspective based 3D pose reconstruction

**

3D reconstruction

Driven by recent computer vision applications, recovering 3D pose in the field of figure skating has become increasingly important. This work proposes a multitask architecture based on a calibrated multi-camera system to facilitate jointly 3D jump pose of figure skater in the presence of the 2D Confidence Map. Since 2D pose is inherently ambiguous as more than one 3D poses can map to a same 2D pose, estimating 3D human pose in sports has attracted academic interests for its vast potential recently. Therefore, the analysis of jump pose in figure skating shouldn’t be ignored, as it plays a significant role for figure skater’s behavior understanding. It can not only objectively evaluate the performance of figure skater’s jump for the purpose of sports training and intelligent scoring, but also enhance the audience’s entertainment experience by displaying jump height, spatial trajectory, motion details and other information based on the captured figure skating videos.

demo

one convolution jump
图片:
two convolutions jump

three convolutions jump

2D results