Can I read SHERF: Generalizable Human NeRF from a Single Image on EtoBox?

SHERF: Generalizable Human NeRF from a Single Image by Hu, Shoukang; Hong, Fangzhou; Pan, Liang; Mei, Haiyi; Yang, Lei; Liu, Ziwei is a scholarly article available to read on EtoBox.

What is SHERF: Generalizable Human NeRF from a Single Image about?

Existing Human NeRF methods for reconstructing 3D humans typically rely on multiple 2D images from multi-view cameras or monocular videos captured from fixed camera views. However, in real-world scenarios, human images are often captured from random camera angles, presenting challenges for high-quality 3D human reconstruction. In this paper, we propose SHERF, the first generalizable Human NeRF model for recovering animatable 3D humans from a single input image. SHERF extracts and encodes 3D human representations in canonical space, enabling rendering and animation from free views and poses. To achieve high-fidelity novel view and pose synthesis, the encoded 3D human representations should capture both global appearance and local fine-grained textures. To this end, we propose a bank of 3D-aware hierarchical features, including global, point-level, and pixel-aligned features, to facilitate informative encoding. Global features enhance the information extracted from the single input image and complement the information missing from the partial 2D observation. Point-level features provide strong clues of 3D human structure, while pixel-aligned features preserve more fine-grained detail

Author

Hu, Shoukang; Hong, Fangzhou; Pan, Liang; Mei, Haiyi; Yang, Lei; Liu, Ziwei

Published

2023

Language

EN