AI-Generated Video Forgery Detection and Authentication

Deep learning has a variety of uses and issues it can solve in the real world, but it also has some limitations. The growing use of AI-Morped Videos is one of the most recent and complicated issues. "AI Morphed Videos," which are digitally manipulated still or moving visuals, are made using deep learning techniques. In an AI Morphed Video, the target’s face is superimposed over the original image so that the altered digital data can be used for online frauds, extortion, pornography, etc. It is getting harder and harder to manually discern between true and false as deep learning develops. Therefore, research and development in the field of AI Morphed Video detection are crucial. An overview of the several AI Morphed Video detection strategies is completed in time for the classification of feature-based, temporal-based, and deep feature-based AI Morphed Video detection. The comparison research is based mostly on the key features used, including the face detection architecture, the deep learning architecture, whether it is video-based or image-based, the dataset used, the frames size, and the dataset size used. Along with the comparison, a semi-supervised GAN architecture is also proposed and built to recognize the AI Morphed Video.