Attention-boosted CNNs for improved facial deepfake detection

In this thesis, our primary objective is to delve into the intricate realm of facial biometric authentication systems, with a keen focus on mitigating the significant hurdles they face, particularly in the realms of liveness detection and deep fake identification. These challenges have emerged as formidable adversaries, posing serious threats to the integrity and reliability of facial recognition technologies. To combat these threats effectively, we propose the development of a robust filtering system leveraging state-of-the-art deep learning architectures. Our approach entails the meticulous design and implementation of advanced algorithms capable of discerning between genuine facial features and deceptive manipulations. Through the judicious utilization of deep learning models, we aim to fortify authentication systems against a diverse spectrum of potential attack vectors, ranging from simple spoofing techniques to sophisticated deep fake algorithms. By integrating cutting-edge methodologies in computer vision and artificial intelligence, we endeavor to enhance the resilience of facial biometric authentication systems to emerging threats.