Recovering Physically Plausible Human-Object Interactions from Monocular Videos

2026-06-03

AI Summary

This work addresses the common physical implausibilities—such as interpenetration and floating—in monocular video-based human-object interaction (HOI) reconstruction. To this end, the authors propose RePHO, a novel approach that first initializes an interaction sequence using kinematic estimation and then refines it through physics-guided reinforcement learning within a simulated environment. RePHO introduces an adaptive sampling strategy and a dual self-update mechanism, effectively integrating kinematic priors with physical constraints to alleviate the challenges of policy training under noisy estimates. Evaluated on two standard HOI benchmarks, RePHO significantly outperforms existing methods, demonstrating particularly notable improvements in metrics assessing physical plausibility.

Abstract

In this paper, we propose RePHO, a method to reconstruct physically plausible human-object interactions (HOI) from monocular videos. While existing kinematic-based approaches produce visually plausible motion, they often result in physically implausible artifacts such as interpenetration and object floating. To overcome these issues, we introduce a physics-guided reconstruction framework. We begin with a kinematic estimate and then refine it by training a policy with reinforcement learning (RL). This policy is optimized to reproduce the interaction in a physics simulator. Because kinematic estimates are typically noisy, naive RL training can fail. Therefore, we propose an adaptive sampling strategy with a dual self-updating mechanism that can identify the frames with the most informative and reliable kinematic reconstruction. Our process progressively improves reconstruction quality and yields physically consistent HOI sequences. We demonstrate our approach on two standard HOI benchmarks and achieve clear improvements in physical plausibility metrics over state-of-the-art methods. Project Page: https://dingbang777.github.io/RePHO/

Problem

• human-object interaction
• physical plausibility
• monocular video
• reconstruction artifacts
• physics-based animation

Innovation

• physics-guided reconstruction
• human-object interaction
• reinforcement learning
• adaptive sampling
• monocular video