An overview of state-of-the-art methods for 3D human pose estimation from monocular video

Human pose estimation (HPE) determines the configuration of human body components from a given input, particularly images and videos. The goal of this thesis is to present a comprehensive review of most commonly approaches used to address this subject, which has been widely explored in computer visi...

ver descrição completa

Detalhes bibliográficos
Autor: Pérez Millar, Camila
Tipo de documento: dissertação
Data de publicação:2023
País:España
Recursos:Universitat Politècnica de Catalunya (UPC)
Repositório:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglês
OAI Identifier:oai:upcommons.upc.edu:2117/401727
Acesso em linha:https://hdl.handle.net/2117/401727
Access Level:Acceso aberto
Palavra-chave:Computer vision
HPE
3D HPE
Human Pose Estimation
Visió per ordinador
Àrees temàtiques de la UPC::Informàtica::Arquitectura de computadors
Descrição
Resumo:Human pose estimation (HPE) determines the configuration of human body components from a given input, particularly images and videos. The goal of this thesis is to present a comprehensive review of most commonly approaches used to address this subject, which has been widely explored in computer vision literature. This essay focuses on the state-of-the-art of methods for computing 3D Human Pose Estimation from monocular video, starting with a comprehensive review of the most recent deep learning-based solutions used for 2D and 3D pose estimation, delving into the main 3D HPE techniques and finally carrying out a qualitative comparison of the latest and most widely used models on benchmark datasets.