Please use this identifier to cite or link to this item: https://scholarbank.nus.edu.sg/handle/10635/153741
Title: DEEP LEARNING FOR HUMAN-CENTRIC IMAGE ANALYSIS
Authors: ZHAO JIAN
ORCID iD:   orcid.org/0000-0002-3508-756X
Keywords: Unconstrained Face Recognition, Pose-Invariant Face Recognition, Age-Invariant Face Recognition, Instance-Agnostic Human Parsing, Multi-Human Parsing
Issue Date: 20-Nov-2018
Citation: ZHAO JIAN (2018-11-20). DEEP LEARNING FOR HUMAN-CENTRIC IMAGE ANALYSIS. ScholarBank@NUS Repository.
Abstract: Though great progresses have been made in recent years, the performance of human-centric image analysis in real-world scenarios is still far from being satisfactory. In this thesis, we systematically investigate the problem of human-centric analysis based on deep learning, from face recognition for exploring the identity information to human parsing for exploring the fine-grained semantic information. Pose and age variations are the most challenging factors for face recognition in unconstrained scenarios. We start by designing pose-invariant face recognition approaches, which aim to address unconstrained face recognition with large/extreme pose variations, by synthesizing photorealistic faces with different poses and learning pose-invariant facial representations. Thereafter, we propose age-invariant face recognition algorithms, which elegantly learn disentangled facial representations and photorealistic cross-age faces to solve cross-age face recognition. To enable more detailed human-centric analysis, we then introduce a novel strategy for instance-agnostic human parsing, which effectively aggregates multi-scale contextual information and incorporates human pose based constraints to improve the results. Finally, we propose a novel nested adversarial learning strategy to address the challenging instance-level human parsing (multi-human parsing), and construct a new large-scale and fine-grained benchmark dataset to further push the research frontier of human-centric image analysis and crowded scene understanding.
URI: https://scholarbank.nus.edu.sg/handle/10635/153741
Appears in Collections:Ph.D Theses (Open)

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
01ZhaoJ.pdf628.64 kBAdobe PDF

OPEN

NoneView/Download
02ZhaoJ.pdf8.56 MBAdobe PDF

OPEN

NoneView/Download
03ZhaoJ.pdf22.33 MBAdobe PDF

OPEN

NoneView/Download
04ZhaoJ.pdf25.94 MBAdobe PDF

OPEN

NoneView/Download
05ZhaoJ.pdf28.28 MBAdobe PDF

OPEN

NoneView/Download

Page view(s)

202
checked on Oct 16, 2020

Download(s)

135
checked on Oct 16, 2020

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.