Graduate Studies

 

First Advisor

Benjamin Riggan

Degree Name

Doctor of Philosophy (Ph.D.)

Department

Electrical Engineering

Date of this Version

12-5-2023

Document Type

Dissertation

Comments

Copyright 2023, Cedric Armel Nimpa Fondje

Abstract

Face recognition technology has witnessed significant advancements in recent decades, enabling its widespread adoption in various applications such as security, surveillance, and biometrics applications. However, one of the primary challenges faced by existing face recognition systems is their limited performance when presented with images from different modalities or domains( such as infrared to visible, long range to close range, nighttime to daytime, profile to f rontal, e tc.) Additionally, advancements in camera sensors, analytics beyond the visible spectrum, and the increasing size of cross-modal datasets have led to a particular interest in cross-modal learning for face recognition in the biometrics and computer vision community. Despite a relatively large gap between source and target domains, existing approaches reduce or bridge such domain gaps by either synthesizing face imagery in the target domain using face imagery from the source domain, or by learning cross-modal image representations that are robust to both the source and the target domain. Therefore, this dissertation presents the design and implementation of a novel domain adaptation framework leveraging robust image representations to achieve state-of-the art performance in cross-spectrum and long-range face recognition. The proposed methods use machine learning and deep learning techniques to (1) efficiently ex tract an d le arn do main-invariant embedding from face imagery, (2) learn a mapping from the source to the target domain, and (3) evaluate the proposed framework on several cross-modal face datasets.

Share

COinS