AI-Based Deepfake Video Detection Framework for Forensic Analysis & Digital Evidence Validation

With the rapid advancement of Artificial Intelligence (AI), realistic deepfake videos are being created using deep learning techniques. The manipulated videos can cause significant harm in social, political, and security scenarios. Deepfake videos are created using deep learning techniques to change or replace a person’s facial features, voice, and expressions. The detection of deepfake videos has become challenging with the improvement of deepfake video generation techniques. This project aims to develop a deep learning- based approach to detect deepfake videos using inconsistencies in facial features and patterns. The system integrates Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM). CNN is applied to extract spatial features from each video frame, focusing on facial features, while LSTM is applied to detect abnormal patterns in videos using temporal dependencies between frames. The model is trained and tested using the Celeb-DF dataset to recognize and classify real and deepfake videos accurately.