Probablistic Restoration with Adaptive Noise Sampling for 3D Human Pose Estimation
Abstract
PRPose improves 3D human pose estimation by fitting the hidden probability distribution of the 2D-to-3D lifting process and using adaptive noise sampling to generate plausible multi-hypothesis 3D poses.
Type
Publication
IEEE International Conference on Multimedia and Expo
PRPose can be integrated with lightweight single-hypothesis 3D pose models to generate reasonable multi-hypothesis outputs.
Authors
Ph.D. Student at Zhejiang University
I am a Ph.D. student in the College of Computer Science and Technology at Zhejiang University. My research focuses on spatial intelligence, 3D-AIGC, multi-agent systems, latent reasoning for VLMs, and contrastive learning, with a broader interest in building intelligent systems that connect perception, reasoning, and controllable creation in the world.
Email: haoqin@zju.edu.cn