Deep Fake Detection: Unmasking the Illusion using CNN and LSTM
V Niranjani, Aishwarya S S, T Devamitra, B Jagapreetha · 2023
Concerns regarding the veracity of digital media have arisen due to the development of deepfake technology. In order to effectively combat this new threat, this study offers a novel deepfake detection strategy that combines a number of techniques, including Photoplethysmography (PPG). For the detection and classification of deepfakes, our solution combines PPG with sophisticated deep learning techniques. PPG records physiological signals, enhancing analysis of images and sounds. Convolutional neural networks (CNNs), long short-term memory (LSTM) networks, and audio fingerprinting are used to extract features from a large dataset that was used to train the system. A thorough study showing our multi-algorithm fusion approach’s improved performance in deepfake detection over single-algorithm approaches, including measures like accuracy and precision. This study also explores its resilience against adversarial attacks and consider ethical implications. This research work represents a significant advancement in deepfake detection, emphasizing the integration of various algorithms, including PPG, for improved accuracy and resilience. It has applications in media forensics, content verification, and online security, offering a comprehensive solution to the deepfake detection challenge while promoting responsible digital content authentication.