SiMA-Hand: Boosting 3D Hand-Mesh Reconstruction by Single-to-Multi-View Adaptation

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Wang, Yinqiao, Xu, Hao, Heng, Pheng-Ann, Fu, Chi-Wing
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866911769728909312
author Wang, Yinqiao
Xu, Hao
Heng, Pheng-Ann
Fu, Chi-Wing
author_facet Wang, Yinqiao
Xu, Hao
Heng, Pheng-Ann
Fu, Chi-Wing
contents Estimating 3D hand mesh from RGB images is a longstanding track, in which occlusion is one of the most challenging problems. Existing attempts towards this task often fail when the occlusion dominates the image space. In this paper, we propose SiMA-Hand, aiming to boost the mesh reconstruction performance by Single-to-Multi-view Adaptation. First, we design a multi-view hand reconstructor to fuse information across multiple views by holistically adopting feature fusion at image, joint, and vertex levels. Then, we introduce a single-view hand reconstructor equipped with SiMA. Though taking only one view as input at inference, the shape and orientation features in the single-view reconstructor can be enriched by learning non-occluded knowledge from the extra views at training, enhancing the reconstruction precision on the occluded regions. We conduct experiments on the Dex-YCB and HanCo benchmarks with challenging object- and self-caused occlusion cases, manifesting that SiMA-Hand consistently achieves superior performance over the state of the arts. Code will be released on https://github.com/JoyboyWang/SiMA-Hand Pytorch.
format Preprint
id arxiv_https___arxiv_org_abs_2402_01389
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle SiMA-Hand: Boosting 3D Hand-Mesh Reconstruction by Single-to-Multi-View Adaptation
Wang, Yinqiao
Xu, Hao
Heng, Pheng-Ann
Fu, Chi-Wing
Computer Vision and Pattern Recognition
Estimating 3D hand mesh from RGB images is a longstanding track, in which occlusion is one of the most challenging problems. Existing attempts towards this task often fail when the occlusion dominates the image space. In this paper, we propose SiMA-Hand, aiming to boost the mesh reconstruction performance by Single-to-Multi-view Adaptation. First, we design a multi-view hand reconstructor to fuse information across multiple views by holistically adopting feature fusion at image, joint, and vertex levels. Then, we introduce a single-view hand reconstructor equipped with SiMA. Though taking only one view as input at inference, the shape and orientation features in the single-view reconstructor can be enriched by learning non-occluded knowledge from the extra views at training, enhancing the reconstruction precision on the occluded regions. We conduct experiments on the Dex-YCB and HanCo benchmarks with challenging object- and self-caused occlusion cases, manifesting that SiMA-Hand consistently achieves superior performance over the state of the arts. Code will be released on https://github.com/JoyboyWang/SiMA-Hand Pytorch.
title SiMA-Hand: Boosting 3D Hand-Mesh Reconstruction by Single-to-Multi-View Adaptation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2402.01389