This project creates and detects deepfakes. We use InsightFace for face analysis and imswaper_128.onnx for face-swapping. For detection, CNN and Meso4 models identify deepfakes, supporting secure and reliable digital media analysis. The project has both academic and practical applications.

This project focuses on the creation and detection of deepfakes using advanced AI techniques. Deepfakes, which involve manipulating digital media to create realistic yet fake content, pose significant risks in various domains, including security, privacy, and misinformation. Our project aims to address these challenges by combining powerful face-swapping and detection models.

We leverage InsightFace for precise face analysis, enabling the accurate mapping of facial features. The imswaper_128.onnx model is then used to perform realistic face-swapping, allowing us to generate convincing deepfake videos. This generation process demonstrates the potential misuse of AI technology in creating deceptive media.

To mitigate the risks associated with deepfakes, we have implemented a detection module. This module utilizes Convolutional Neural Networks (CNN) alongside the Meso4 model, which are both trained to detect subtle anomalies and inconsistencies in deepfake content. Our detection system ensures the integrity and authenticity of digital media, making it a valuable tool for cybersecurity and media verification.

This project has both academic and practical applications. Academically, it provides a foundation for further research in deepfake detection and generation, contributing to the development of new methods for handling synthetic media. Practically, the system can be employed by law enforcement, media outlets, and cybersecurity professionals to identify and combat malicious deepfakes.