Bilkent University
Department of Computer Engineering

---

CS590 SEMINAR

 

FACE MANIPULATION DETECTION USING ONE CLASS CLASSIFIERS

 

Hasan Şerefal
MS Student
(Supervisor: Asst. Prof. Dr. Shervin R. Arashloo)
Computer Engineering Department
Bilkent University

The rapid progress in deep learning is affecting so many areas nowadays. Computer Vision is one of the disciplines affected by this progress seriously. Synthetic image generation comes to a point such that generated images that can not be distinguished from the real one by human vision are achievable. This generated images creates important concerns in so many areas. This synthetic image generation can also be used to create face manipulation purposes. With swapping faces of the people, they can be presented like they are saying things they don’t actually say or doing things they actually don’t. With this fake presentation, people can be defamed, or targeted. Due to reasons stated above, detection of the manipulated faces is a really important task. In this task we are trying to detect these manipulated faces using One Class Classifier with features extracted from Convolutional Neural Network.

 

DATE: 26 April 2021, Monday @ 15:30