Bézier Degradation Modeling for LiDAR-based Human Motion Capture

2026-05-19

AI Summary

This work addresses the challenges of pose jitter and failure in LiDAR-based human motion capture caused by unstable inputs and severe occlusions. To this end, the authors propose BMLiCap, a novel framework that employs a coarse-to-fine strategy featuring trajectory-preserving Bézier curves for temporally compressible motion modeling. The framework further introduces a Time-scale Motion Transformer (TMT) and a Multi-level Motion Aggregator (MMA) to enable multi-scale temporal modeling and adaptive feature fusion. Evaluated on four benchmarks, including LiDARHuman26M, BMLiCap significantly enhances the stability and fidelity of motion reconstruction, effectively suppresses prediction jitter, and accurately recovers missing poses under heavy occlusion, achieving state-of-the-art accuracy and temporal consistency.

Abstract

LiDAR-based 3D human motion capture has broad applications in fields such as autonomous driving and robotics, where accurate motion reconstruction is crucial. However, existing methods often struggle with unstable inputs and severe occlusions, leading to jittery or even failed pose predictions. To address these challenges, we propose BMLiCap, a coarse-to-fine framework that models motion using temporally compressible Bézier curves. By reducing control points through a trajectory-preserving strategy, we obtain a coherent and learning-friendly motion representation. To reconstruct human actions from LiDAR point-cloud cues, we design a progressive motion-reconstruction module. Specifically, a Time-scale Motion Transformer (TMT) is introduced to predict motion curves at multiple temporal scales, and a Multi-level Motion Aggregator (MMA) is utilized to adaptively fuse the multi-scale curves to recover detailed, temporally coherent poses, effectively bridging observation gaps caused by occlusions and noise. Across four mainstream benchmarks LiDARHuman26M, FreeMotion, NoiseMotion, and SLOPER4D, BMLiCap achieves state-of-the-art accuracy and temporal continuity in complex scenes, demonstrating its ability to compensate for severe occlusions and reduce prediction jitter.

Problem

• LiDAR-based human motion capture
• occlusions
• motion jitter
• unstable inputs
• pose prediction

Innovation

• Bézier curves
• LiDAR-based motion capture
• temporal coherence
• multi-scale motion reconstruction
• occlusion robustness