The content introduces the U-HMR framework for human mesh recovery from arbitrary multi-view images. It discusses the challenges of estimating camera poses and recovering human mesh, proposing a decoupled structure for efficient processing. The CBD structure splits tasks into camera pose estimation (CPE) and view fusion (AVF). Extensive experiments on datasets validate the efficacy of the proposed components.
To Another Language
from source content
arxiv.org
Thông tin chi tiết chính được chắt lọc từ
by Xiaoben Li,M... lúc arxiv.org 03-20-2024
https://arxiv.org/pdf/2403.12434.pdfYêu cầu sâu hơn