Date of Award
12-2024
Document Type
Thesis
Degree Name
Master of Science (MS)
Department
Electrical Engineering and Computer Science
First Advisor
Michael C. King, Ph.D.
Second Advisor
Siddhartha Bhattacharyya, Ph.D.
Third Advisor
Jignya M. Patel, Ph.D.
Fourth Advisor
Brian A. Lail, Ph.D.
Abstract
In the rapidly evolving digital age, biometric authentication systems have grown tremendously in importance. Traditional facial recognition systems are commonly implemented in current businesses due to their simplicity and efficiency. However, both systems are increasingly vulnerable to presentation attacks and other advanced spoofing techniques. The benefits of the easy application of fixed architecture in legacy systems lead to authorized users being vulnerable to spoof attacks involving 3D masks, video playbacks, and pilfered pictures to gain unauthorized access. The security gap indicates the fundamentally limiting aspect of all static solutions, which cannot verify liveliness or an individual’s presence in real time. For better security in biometric systems, our thesis proposes a dynamic facial expression detection mechanism integrated into an existing face recognition framework. Real-time facial expressions add an extra layer to the identification process to counter ordinary vulnerabilities like photo and deep fake attacks. The more technological the solutions are, the worse exploitation by malicious attacks is, so our system is designed to provide protection. We conjecture that live human expressions are dynamic and complex, making imitations challenging. We examined the capabilities of DeepFace, an open-source computer vision tool designed to facilitate behavioral analysis by the affect computing research community on the Affective Behavior Analysis in the Wild (ABAW) dataset, was utilized to detect which facial expressions appeared in a video sequence. While FaceReader had higher overall accuracy, DeepFace was implemented to demonstrate the design prototype to fuse expression detection with the accessibility of the application programmer interface.
Recommended Citation
Cheeti, Upanya, "Remote Authentication Using Facial Recognition and Expressions" (2024). Theses and Dissertations. 1484.
https://repository.fit.edu/etd/1484