Deepfake Detection
Deepfake is an image or video-based form of synthetic media. It includes changing the emotions of a person or swapping his face with another individual replacing the identity. Over the past few years, with the advance in technology deepfakes are being used for unethical uses. And it can be foreseen that the situation will get worse with time. So, it's a need of the hour to devise robust models for detecting deepfakes. Major misuses of deepfakes that are the topic of research of some scientists include cyber threats, incitement to violence, dissemination of fake news, using deepfakes for blackmailing, intimidation, sabotage, and ideological manipulation. This project is about solving the problem of widespread deepfakes that are hard to identify as fake. From the research surveys, it is clear that the problem of face manipulation requires generalized solutions. Among the promising approaches to improve the generalization problem of current face manipulation detectors is one-class learning models, commonly used in the anomaly and novelty detection literature. Such models learn from real examples and do not require examples of manipulated data to train fake detectors. As a result, they are expected to generalize better for the detection of novel (unseen) face manipulation techniques. Of course, such detection techniques come with their own set of problems that range from data representation and model design to learning objectives, among others. I am working on implementing a few shot model for anomaly detection to detect deepfakes. I am also currently working on detecting deepfakes by using yoloface and facenet models and then applying the CNNs to test the accuracy of such model which extracts the face from the whole image and then from the patterns in the image it tells whether the image is real or fake. The reason for testing different models is to find a more optimized and generalized solution to the deepfake problem.
Keywords: Machine Learning,Deep Learning,Computer Vision,HTDG Model,YoloFace Model,FaceNet Model
Tools: Python ,Pytorch,PyCharm
Department: Department of Computer Science
Project Team Members
| Name |
Email |
|
Fateh Latif
|
iamfateh01@gmail.com |
Project Poster
