Reconstructing Global Multi-Person 3D Motions From A Single Moving Camera
Date
2025-04-15
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
The reconstruction of 3D human models from in-the-wild videos has emerged as a highly promising research direction, with increasingly practical and widespread applications. To address the challenges of depth ambiguity and human occlusions in monocular videos, an optimization-based framework is proposed that combines SLAM techniques with human motion priors to recover global multi-person motion from a single moving camera. Camera parameters are estimated using SLAM, while a robust tracking algorithm is adopted to mitigate occlusion effects. A learned motion smoothness prior is further integrated into the optimization process. The SMPL model and camera parameters are jointly optimized to align the reconstructed 3D poses with 2D detections. Experimental results on the 3DPW dataset demonstrate that the proposed method outperforms existing approaches in both accuracy and robustness.
Description
To address 3D human reconstruction in multiple dynamic scenarios, a new optimization-based framework, Opti-Pose3D, is proposed to recover global multi-person motion from a single moving camera by integrating SLAM techniques with a learned human motion prior. To detect and associate individuals across frames, a robust tracking algorithm is employed to handle occlusions effectively. In addition, a learned motion smoothness prior is incorporated into the optimization pipeline to enhance temporal consistency. Using the initial camera parameters estimated by SLAM, human motion trajectories are reconstructed by jointly optimizing the SMPL model and camera parameters to align the projected 3D poses with the detected 2D observations.
Keywords
Reconstruction, SLAM, SmoothNet