The concept for this project is to let user verify the authencity of an image online or in their computer. This will allow to crackdown on the propoganda materials or fake information online especially in social media. Initially, the 4th industrial revolution that is changing the way we access information has bought many challenges for the users to tackle. At times, an individual is left tangled between a post or information that is effectively using deepfake algorithm to inflict harm to their opponent's image. Now the art of deepfake is that It can't be spotted with naked eyes if the data is real or fake. Thus, our solution is here to help the social media platforms, electronic and digital media forums or netizens alike to help differentiate propoganda and real to stay informed and escape the world of delusion.
You can find the trained model here - https://drive.google.com/file/d/1BXTHhR_VDIGKj2BJ6KjqaB4-hDpHP8w4/view?usp=sharing. This model was trained on the FaceForensics, Kaggle, and the CELEB-DF datasets to achieve an accuracy of 92% across videos with over 100 frames.