Learning to Estimate 3D Hand Pose from Single RGB Images

TLDR

Low‑cost consumer depth cameras and deep learning have enabled reasonable 3D hand pose estimation from single depth images, but the task is far more ambiguous without depth information. The paper proposes a deep network that learns a network‑implicit 3D articulation prior to estimate 3D hand pose from regular RGB images. The method uses this deep network and is trained on a large synthetic 3D hand pose dataset. The network, combined with image‑detected keypoints, yields accurate 3D pose estimates, and experiments on diverse test sets—including sign‑language recognition—demonstrate its feasibility on single RGB images.

Abstract

Low-cost consumer depth cameras and deep learning have enabled reasonable 3D hand pose estimation from single depth images. In this paper, we present an approach that estimates 3D hand pose from regular RGB images. This task has far more ambiguities due to the missing depth information. To this end, we propose a deep network that learns a network-implicit 3D articulation prior. Together with detected keypoints in the images, this network yields good estimates of the 3D pose. We introduce a large scale 3D hand pose dataset based on synthetic hand models for training the involved networks. Experiments on a variety of test sets, including one on sign language recognition, demonstrate the feasibility of 3D hand pose estimation on single color images.