3D human pose estimation from a single image via exemplar augmentation

Abstract 3D human pose estimation from a single image is a challenging problem due to occlusion, viewpoint variance, and the ill-posed nature of back projection. We follow a standard two-step pipeline which first detects 2D joint locations and uses them to infer 3D pose. For the first step, we use a recent deep learning-based detector. For the second step, we propose a novel exemplar-based algorithm to implicitly augment the exemplar set for 3D human pose estimation. The motivation of this algorithm is to well represent various poses in the real world with finite real exemplars. We achieve it by a strategy of synthesizing virtual candidate poses which ensures that the augmented exemplar set has much more variety. Moreover, we also present an effective approach to select the best exemplar from candidate set to well match the detected 2D pose. Experimental results show that our method achieves competitive performance on Human3.6M dataset.