Technology has advanced so much that anyone can edit images with the software available and converts the real picture into fake picture. Selfies have become an integral part of photography in recent years, and they are even considered a powerful and trustworthy medium of communication. To identify fake faces in various systems, many spoof detection techniques have been created, but face forgery continues as a challenge in social media platform. This paper specifies a robust algorithm that can detect fake faces in shared photographs on social media. A Haar cascade classifier is used for feature selection and CNN for classification whether the input image is real or fake. This paper will analyze the existing system with the current accuracy and how our method reaches the highest accuracy for deepfake detection.

Haar Cascade Classifier, Convolutional Neural Networks, Face Spoofing