HandNeRF: Learning to Reconstruct Hand-Object Interaction Scene from a Single RGB Image

TL;DR

HandNeRF is a novel implicit function that reconstructs 3D hand-object scenes from a single RGB image by leveraging hand shape constraints, improving accuracy over existing methods and aiding robotic manipulation tasks.

Contribution

This work introduces HandNeRF, a generalizable implicit model that encodes hand-object correlations for improved 3D scene reconstruction from minimal input.

Findings

Outperforms existing methods in reconstructing novel grasp configurations

Enables more accurate downstream robotic tasks like grasping and motion planning

Demonstrates robustness on real-world datasets

Abstract

This paper presents a method to learn hand-object interaction prior for reconstructing a 3D hand-object scene from a single RGB image. The inference as well as training-data generation for 3D hand-object scene reconstruction is challenging due to the depth ambiguity of a single image and occlusions by the hand and object. We turn this challenge into an opportunity by utilizing the hand shape to constrain the possible relative configuration of the hand and object geometry. We design a generalizable implicit function, HandNeRF, that explicitly encodes the correlation of the 3D hand shape features and 2D object features to predict the hand and object scene geometry. With experiments on real-world datasets, we show that HandNeRF is able to reconstruct hand-object scenes of novel grasp configurations more accurately than comparable methods. Moreover, we demonstrate that object reconstruction from HandNeRF ensures more accurate execution of downstream tasks, such as grasping and motion planning for robotic hand-over and manipulation. Homepage: https://samsunglabs.github.io/HandNeRF-project-page/