ReFlow: Self-correction Motion Learning for Dynamic Scene Reconstruction

2026-04-01

AI Summary

Monocular dynamic scene reconstruction often suffers from instability and motion estimation bias due to incomplete initialization of dynamic regions, and reliance on external dense motion priors can introduce additional errors. This work proposes ReFlow, a unified framework that introduces a novel self-correcting optical flow matching mechanism to disentangle static and dynamic scene components without external supervision while enforcing multi-view consistency. By constructing a complete canonical space, employing decoupled dynamic modeling, and jointly optimizing neural radiance fields with optical flow correspondences, ReFlow establishes a new paradigm for monocular 4D reconstruction. Experiments demonstrate that the method significantly outperforms existing approaches across diverse dynamic scenes, achieving superior reconstruction accuracy and robustness.

Abstract

We present ReFlow, a unified framework for monocular dynamic scene reconstruction that learns 3D motion in a novel self-correction manner from raw video. Existing methods often suffer from incomplete scene initialization for dynamic regions, leading to unstable reconstruction and motion estimation, which often resorts to external dense motion guidance such as pre-computed optical flow to further stabilize and constrain the reconstruction of dynamic components. However, this introduces additional complexity and potential error propagation. To address these issues, ReFlow integrates a Complete Canonical Space Construction module for enhanced initialization of both static and dynamic regions, and a Separation-Based Dynamic Scene Modeling module that decouples static and dynamic components for targeted motion supervision. The core of ReFlow is a novel self-correction flow matching mechanism, consisting of Full Flow Matching to align 3D scene flow with time-varying 2D observations, and Camera Flow Matching to enforce multi-view consistency for static objects. Together, these modules enable robust and accurate dynamic scene reconstruction. Extensive experiments across diverse scenarios demonstrate that ReFlow achieves superior reconstruction quality and robustness, establishing a novel self-correction paradigm for monocular 4D reconstruction.

Problem

• dynamic scene reconstruction
• monocular 4D reconstruction
• incomplete initialization
• motion estimation
• error propagation

Innovation

• self-correction
• dynamic scene reconstruction
• monocular 4D reconstruction
• flow matching
• canonical space