Efficient Prediction of Real vs Fake Face using Transfer Learning and Deep Learning

Artificial Intelligence introduced for human activity recognition in real time. This technology is used by many people. Fake faces are used by cyber hackers for money debit and other suspicious activity. Government need to monitor more on those who creates fake face for suspicious activity. Recently fake faces create several issues in society and for example social media, theft in society. Accurate Fake face detection is necessary for eliminating issues created by fake faces in society. Deep learning is a model to predict real faces and fake faces. Transfer learning is highly accurate in prediction over deep learning. ResNet50, MobileNetv2 from transfer learning and Convolution Neural Network(CNN ) from deep learning are used in proposed research. ResNet50 transfer learning model shows 96% accuracy over other transfer learning model MobileNetv2 to predict real and fake faces. Proposed transfer learning models are highly efficient to detect real and fake faces over deep learning models such as CNN.