A Study On Deep Fakes Detection Using Blinks and Tracker
C Vasanthakumar, S Karthikeyan, V R Raghunandhan, S Sanjay, R Tushar · 2022
Deep learning, a type of artificial intelligence (AI), replicates how the human brain processes data when processing data for tasks including speech recognition, object detection, visual object recognition, language translation, and decision-making. Generative Adversarial Network (GAN) is an outstanding generative version which can be extensively utilized in diverse applications. Recent research has indicated that it's far feasible to achieve faux face photographs with an excessive visible excellence primarily based totally in this novel version. For instance, the example may be impacted by a person's orientation, age, the hour of the day, or level of personal wellbeing. The misuse of the fake faces in photo manipulation could lead to certain ethical, moral, and legal issues. Therefore, in this work we first recommend a Convolution Neural Network (CNN) based entirely technique to find fake face photos produced by the modern-day pleasant technique, and we provide practical evidences to indicate that the proposed technique can create high-satisfactory results with a median accuracy over 87.5%which was accomplished in previous papers and projects. In order to further support the logic of our method, we also provide comparison results assessed on a few variations of the suggested CNN architecture, including the excessive by skip filter, the variety of the layer agencies, and the activation function.