DeepDetector is a deep learning network designed to detect and recognize manipulated faces in images and videos, including deepfakes. This artificial neural network can distinguish between genuine images and computer-generated forgeries by analyzing thousands of real and deepfake images. The technology can extract visible faces in a picture or video, analyze them, and detect deepfake traces with an accuracy rate of approximately 93%. The output of the analysis includes the probability of the input being a deepfake and the Activation Map, which offers an explanation behind the software's decision by substantiating the classification.DeepDetector can not only identify FaceSwaps and other AI-manipulations but also analyze characteristics of AI-generated content to detect synthetic media and AI-generated deepfakes. The tool also offers cloud-based access to customizable APIs to integrate DeepDetector into user workflows, ensuring data protection and privacy through compliance with European laws and regulations regarding data protection, privacy, and responsible AI.DeepDetector can be used for compliance purposes such as KYC, video conferencing, and journalism, as well as in penetration testing. The tool is designed to provide explainable and accurate results, with 10+ deepfake types detected and an image analysis time of 1 second. In summary, DeepDetector is a powerful and reliable tool for detecting deepfakes and other AI-manipulations in images and videos.
