Academic Journal

Body Size and Depth Disambiguation in Multi-Person Reconstruction from Single Images

Bibliographic Details
Title: Body Size and Depth Disambiguation in Multi-Person Reconstruction from Single Images
Authors: Ugrinovic Kehdy, Nicolas, Ruiz Ovejero, Adrià, Agudo Martínez, Antonio, Sanfeliu Cortés, Alberto, Moreno-Noguer, Francesc
Contributors: Ministerio de Ciencia, Innovación y Universidades (España), Agencia Estatal de Investigación (España), Ministerio de Economía y Competitividad (España), Consejo Superior de Investigaciones Científicas [https://ror.org/02gfc7t72]
Source: UPCommons. Portal del coneixement obert de la UPC
Universitat Politècnica de Catalunya (UPC)
Digital.CSIC. Repositorio Institucional del CSIC
Consejo Superior de Investigaciones Científicas (CSIC)
instname
Publication Status: Preprint
Publisher Information: IEEE, 2021.
Publication Year: 2021
Subject Terms: Optimization, FOS: Computer and information sciences, Image processing--Digital techniques, Gradient descent, Àrees temàtiques de la UPC::Informàtica::Robòtica, Computer Vision and Pattern Recognition (cs.CV), SMPL, Computer Science - Computer Vision and Pattern Recognition, 02 engineering and technology, 3D pose and shape estimation, 03 medical and health sciences, 0302 clinical medicine, Pattern recognition, Pose and shape, 0202 electrical engineering, electronic engineering, information engineering, Imatges--Processament--Tècniques digitals, Computer vision, Informàtica::Robòtica [Àrees temàtiques de la UPC], Multi-person 3D pose estimation
Description: We address the problem of multi-person 3D body pose and shape estimation from a single image. While this problem can be addressed by applying single-person approaches multiple times for the same scene, recent works have shown the advantages of building upon deep architectures that simultaneously reason about all people in the scene in a holistic manner by enforcing, e.g., depth order constraints or minimizing interpenetration among reconstructed bodies. However, existing approaches are still unable to capture the size variability of people caused by the inherent body scale and depth ambiguity. In this work, we tackle this challenge by devising a novel optimization scheme that learns the appropriate body scale and relative camera pose, by enforcing the feet of all people to remain on the ground floor. A thorough evaluation on MuPoTS-3D and 3DPW datasets demonstrates that our approach is able to robustly estimate the body translation and shape of multiple people while retrieving their spatial arrangement, consistently improving current state-of-the-art, especially in scenes with people of very different heights
Document Type: Article
Conference object
File Description: application/pdf
DOI: 10.1109/3dv53792.2021.00016
DOI: 10.48550/arxiv.2111.01884
DOI: 10.13039/501100003329
DOI: 10.13039/501100011033
Access URL: http://arxiv.org/pdf/2111.01884
http://arxiv.org/abs/2111.01884
http://hdl.handle.net/10261/263182
Rights: IEEE Copyright
CC BY NC ND
Accession Number: edsair.doi.dedup.....0625c6034a92702100bb16f3005cb953
Database: OpenAIRE
Description
DOI:10.1109/3dv53792.2021.00016