HOLD: Category-agnostic 3D Reconstruction of Interacting Hands and Objects from Video

TL;DR

HOLD is a novel category-agnostic method that jointly reconstructs 3D hands and objects from monocular videos without relying on 3D annotations, outperforming supervised methods in diverse settings.

Contribution

The paper introduces HOLD, the first category-agnostic approach for joint 3D reconstruction of hands and objects from monocular videos without needing 3D annotations.

Findings

Outperforms supervised baselines in lab and wild settings

Does not rely on 3D hand-object annotations

Robustly reconstructs from in-the-wild videos

Abstract

Since humans interact with diverse objects every day, the holistic 3D capture of these interactions is important to understand and model human behaviour. However, most existing methods for hand-object reconstruction from RGB either assume pre-scanned object templates or heavily rely on limited 3D hand-object data, restricting their ability to scale and generalize to more unconstrained interaction settings. To this end, we introduce HOLD -- the first category-agnostic method that reconstructs an articulated hand and object jointly from a monocular interaction video. We develop a compositional articulated implicit model that can reconstruct disentangled 3D hand and object from 2D images. We also further incorporate hand-object constraints to improve hand-object poses and consequently the reconstruction quality. Our method does not rely on 3D hand-object annotations while outperforming fully-supervised baselines in both in-the-lab and challenging in-the-wild settings. Moreover, we qualitatively show its robustness in reconstructing from in-the-wild videos. Code: https://github.com/zc-alexfan/hold