4D-HOF: Hand-Object Flow Matching for Feed-Forward 4D Interaction Reconstruction

2026-10-06

AI Summary

This study addresses the instability of existing 4D hand-object interaction reconstruction methods, which typically rely on costly optimization or noise-prone generative processes. To overcome these limitations, this work proposes a post-processing-free feedforward framework that first leverages vision foundation models for coarse-grained estimation and subsequently performs 4D interaction reconstruction via conditional flow matching. The core innovation lies in directly injecting physical constraints and 2D evidence guidance during the generative transport process. By integrating these priors into the flow-matching trajectory, the proposed method achieves state-of-the-art performance on out-of-distribution benchmarks, enabling stable and precise 4D hand-object interaction reconstruction in unconstrained in-the-wild scenarios.

Abstract

Existing methods for 4D hand-object reconstruction often rely on costly per-sequence optimization, while generative approaches typically synthesize interactions from random noise, which can lead to unstable interaction prediction. We introduce 4D-HOF, a feed-forward framework that reconstructs 4D hand-object interactions from coarse but informative estimates produced by vision foundation models. Concretely, we learn a conditional flow matching model that transports foundation-model-derived hand-object states toward an interaction manifold, allowing the model to correct errors in translation, rotation, and alignment in a feed-forward manner. A key advantage of our generative formulation is that it naturally enables test-time guidance within the transport process. Rather than applying a separate post-hoc optimization after reconstruction, we directly steer the evolving generative states using physical interaction constraints and observed 2D evidence, allowing the reconstruction to be refined as part of the generative process itself. By training the generative model on diverse datasets, 4D-HOF generalizes robustly to challenging in-the-wild scenarios. Experiments on out-of-domain benchmarks show that 4D-HOF achieves state-of-the-art performance, producing more stable and accurate 4D hand-object reconstructions.

Problem

• 4D hand-object reconstruction
• per-sequence optimization
• generative modeling
• interaction prediction stability

Innovation

• 4D Hand-Object Reconstruction
• Flow Matching
• Feed-Forward Framework
• Test-Time Guidance
• Vision Foundation Models