VIBE: Video Inference for Human Body Pose and Shape Estimation
This addresses the lack of ground-truth 3D motion data for training in computer vision, enabling more accurate human motion analysis for applications like behavior understanding.
The paper tackles the problem of inaccurate and unnatural motion sequences in video-based 3D human pose and shape estimation by proposing VIBE, which uses adversarial learning with a motion capture dataset and unpaired 2D keypoints, achieving state-of-the-art performance on challenging datasets.
Human motion is fundamental to understanding behavior. Despite progress on single-image 3D pose and shape estimation, existing video-based state-of-the-art methods fail to produce accurate and natural motion sequences due to a lack of ground-truth 3D motion data for training. To address this problem, we propose Video Inference for Body Pose and Shape Estimation (VIBE), which makes use of an existing large-scale motion capture dataset (AMASS) together with unpaired, in-the-wild, 2D keypoint annotations. Our key novelty is an adversarial learning framework that leverages AMASS to discriminate between real human motions and those produced by our temporal pose and shape regression networks. We define a temporal network architecture and show that adversarial training, at the sequence level, produces kinematically plausible motion sequences without in-the-wild ground-truth 3D labels. We perform extensive experimentation to analyze the importance of motion and demonstrate the effectiveness of VIBE on challenging 3D pose estimation datasets, achieving state-of-the-art performance. Code and pretrained models are available at https://github.com/mkocabas/VIBE.