Monocular 3D face reconstruction

Hoda Mohaghegh

doi:10.26182/re56-t090

Monocular 3D face reconstruction

Hoda Mohaghegh

School of Physics, Mathematics and Computing

Research output: Thesis › Doctoral Thesis

56 Downloads (Pure)

Abstract

Monocular 3D face reconstruction is a significant research area within computer vision and graphics. Reconstructed 3D faces have wide applications in areas like face recognition, human-computer interaction, virtual/augmented reality, security, and healthcare. Despite remarkable strides in deep learning-based methods, accurately reconstructing a high-fidelity 3D face from a single image remains challenging due to issues such as occlusions, varying poses, and complex lighting conditions. Existing approaches are typically limited to near-frontal images that are free of occlusions and struggle to handle challenging conditions such as occlusions and out-of-plane rotations.

Original language	English
Qualification	Doctor of Philosophy
Awarding Institution	The University of Western Australia
Supervisors/Advisors	Bennamoun, Mohammed, Supervisor Boussaid, Farid, Supervisor Laga, Hamid, Supervisor, External person Rahmani, Hossein, Supervisor, External person
Award date	12 Jun 2024
DOIs	https://doi.org/10.26182/re56-t090
Publication status	Unpublished - 2023

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.26182/re56-t090

THESIS - DOCTOR OF PHILOSOPHY - MOHAGHEGH Hoda - 2023
This work is protected by Copyright. You may print or download ONE copy of this document for the purpose of your own non-commercial research or study. Any other use requires permission from the copyright owner. The Copyright Act requires you to attribute any copyright works you quote or paraphrase.

1 Conference paper
1 Article

Reinforced Learning for Label-Efficient 3D Face Reconstruction
Mohaghegh, H., Rahmani, H., Laga, H., Boussaid, F. & Bennamoun, M., 2023, Proceedings - ICRA 2023: IEEE International Conference on Robotics and Automation. IEEE, Institute of Electrical and Electronics Engineers, p. 6028-6034 7 p. (Proceedings - IEEE International Conference on Robotics and Automation; vol. 2023-May).
Research output: Chapter in Book/Conference paper › Conference paper › peer-review

Open Access
File
1 Citation (Scopus)

65 Downloads (Pure)
Robust monocular 3D face reconstruction under challenging viewing conditions
Mohaghegh, H., Boussaid, F., Laga, H., Rahmani, H. & Bennamoun, M., 1 Feb 2023, In: Neurocomputing. 520, p. 82-93 12 p.
Research output: Contribution to journal › Article › peer-review
3 Citations (Web of Science)

Cite this

@phdthesis{e3941d77379a45a6a4e2be5f29fd77a2,

title = "Monocular 3D face reconstruction",

abstract = "Monocular 3D face reconstruction is a significant research area within computer vision and graphics. Reconstructed 3D faces have wide applications in areas like face recognition, human-computer interaction, virtual/augmented reality, security, and healthcare. Despite remarkable strides in deep learning-based methods, accurately reconstructing a high-fidelity 3D face from a single image remains challenging due to issues such as occlusions, varying poses, and complex lighting conditions. Existing approaches are typically limited to near-frontal images that are free of occlusions and struggle to handle challenging conditions such as occlusions and out-of-plane rotations.",

keywords = "3D face reconstruction, Self-augmentation, Graph Convolutional Networks (GCN), Active learning, Diffusion models, Label-efficient 3D face reconstruction, Reinforcement learning, Denoising Diffusion Probabilistic Models (DDPM)",

author = "Hoda Mohaghegh",

year = "2023",

doi = "10.26182/re56-t090",

language = "English",

school = "The University of Western Australia",

}

TY - BOOK

T1 - Monocular 3D face reconstruction

AU - Mohaghegh, Hoda

PY - 2023

Y1 - 2023

N2 - Monocular 3D face reconstruction is a significant research area within computer vision and graphics. Reconstructed 3D faces have wide applications in areas like face recognition, human-computer interaction, virtual/augmented reality, security, and healthcare. Despite remarkable strides in deep learning-based methods, accurately reconstructing a high-fidelity 3D face from a single image remains challenging due to issues such as occlusions, varying poses, and complex lighting conditions. Existing approaches are typically limited to near-frontal images that are free of occlusions and struggle to handle challenging conditions such as occlusions and out-of-plane rotations.

AB - Monocular 3D face reconstruction is a significant research area within computer vision and graphics. Reconstructed 3D faces have wide applications in areas like face recognition, human-computer interaction, virtual/augmented reality, security, and healthcare. Despite remarkable strides in deep learning-based methods, accurately reconstructing a high-fidelity 3D face from a single image remains challenging due to issues such as occlusions, varying poses, and complex lighting conditions. Existing approaches are typically limited to near-frontal images that are free of occlusions and struggle to handle challenging conditions such as occlusions and out-of-plane rotations.

KW - 3D face reconstruction

KW - Self-augmentation

KW - Graph Convolutional Networks (GCN)

KW - Active learning

KW - Diffusion models

KW - Label-efficient 3D face reconstruction

KW - Reinforcement learning

KW - Denoising Diffusion Probabilistic Models (DDPM)

U2 - 10.26182/re56-t090

DO - 10.26182/re56-t090

M3 - Doctoral Thesis

ER -

Monocular 3D face reconstruction

Abstract

UN SDGs

Access to Document

Fingerprint

Research output

Reinforced Learning for Label-Efficient 3D Face Reconstruction

Robust monocular 3D face reconstruction under challenging viewing conditions

Cite this