Deep learning-based video forgery detection

• Main Author: Cao, Xinyi
• Other Authors: Alex Chichung Kot
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Computer science and engineering::Computing methodologies::Artificial intelligence; Engineering::Electrical and electronic engineering::Electronic systems::Signal processing
• Online Access: https://hdl.handle.net/10356/157582
• Institution: Nanyang Technological University

The harm of deepfake is becoming more and more serious in today’s new media era, especially in video deepfake. Therefore, we conduct experiments on two public video datasets Celeb-DF-v2, DFDC and a relabelled TMC media dataset, using an end-to-end structure of video input and video classification output, combining the state-of-the-art Convolutional Neural Network (CNN) models with the Vision Transformer architecture and the Long Short-Term Memory (LSTM) architecture. It is found that the longer the frame length of the video, the more accurate the detection. In the case of video length of 30 frames, we obtain competitive AUC scores of 0.932 on the DFDC dataset, 0.980 on the Celeb-DF-V2 dataset and 0.953 on the TMC dataset.