Survey on multi-person 3D reconstruction from monocular view.

Abstract

Reconstructing human pose and shape from a monocular view is a longstanding problem in the field of computer graphics. The goal is to reconstruct the 3D model of human body surface from an image or a video. Most existing monocular human reconstruction methods have achieved great progress which primarily focus on single-person reconstruction. However, research on multi-person reconstruction from a monocular perspective still face certain challenges. In this paper, we comprehensively introduce existing methods in the field of monocular multi-person reconstruction. Additionally, we discuss the challenges and insights into future research directions related to multi-person reconstruction.