Title: U² Net-Plus and background removal-based PIFu-HD: human body reconstruction in complex background

Authors: Guorun Wang; Xudong Liu; Kuo-Yi Lin; Fuhjiun Hwang

Addresses: Tongji University, No. 4800, Caoan Road, JiaDing District, Shanghai, China ' Tongji University, No. 4800, Caoan Road, JiaDing District, Shanghai, China ' Tongji University, No. 4800, Caoan Road, JiaDing District, Shanghai, China ' Shanghai Film Art Academy, No. 188 Dalwen Road, Zhangjiang, Pudong New Area, Shanghai, China

Abstract: At present, algorithms for human motion capture and 3D human reconstruction are not perfect in some experiment, such as PIFu-HD. In addition, there are some reconstruction errors in practical application. We believe that there is background interference. In this paper, we provide the same action in different backgrounds for PIFu-HD model. We endow a single RSU block of U² Net with multi-receptive field mechanism, and use it for dataset of salient object detection (SOD), achieving a significant improvement in accuracy. Then we do the comparative experiment which provides an opportunity of analysing the possible causes of errors in existing algorithms. Thus we propose a new method, a background removal based PIFu-HD. In the end, we use well-constructed dance images and videos for relevant modelling and comparison. We also present the well-constructed dataset and a formulated standard for Cha-Cha, which is of great importance to model-training.

Keywords: image segmentation; 3D human reconstruction; salient object detection; SOD; dancing dataset.

DOI: 10.1504/IJIMS.2022.128640

International Journal of Internet Manufacturing and Services, 2022 Vol.8 No.4, pp.330 - 351

Received: 08 Feb 2022
Accepted: 29 Jul 2022
Published online: 31 Jan 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article