Transformation on Computer-Generated Facial Image to Avoid Detection by Spoofing Detector
Abstract
Making computer-generated (CG) images more difficult to detect is an interesting problem in computer graphics and security. While most approaches focus on the image rendering phase, this paper presents a method based on increasing the naturalness of CG facial images from the perspective of spoofing detectors. The proposed method is implemented using a convolutional neural network (CNN) comprising two autoencoders and a transformer and is trained using a black-box discriminator without gradient information. Over 50% of the transformed CG images were not detected by three state-of-the-art spoofing detectors. This capability raises an alarm regarding the reliability of facial authentication systems, which are becoming widely used in daily life.
Cite
@article{arxiv.1804.04418,
title = {Transformation on Computer-Generated Facial Image to Avoid Detection by Spoofing Detector},
author = {Huy H. Nguyen and Ngoc-Dung T. Tieu and Hoang-Quoc Nguyen-Son and Junichi Yamagishi and Isao Echizen},
journal= {arXiv preprint arXiv:1804.04418},
year = {2018}
}
Comments
Accepted to be Published in Proceedings of the IEEE International Conference on Multimedia and Expo (ICME) 2018, San Diego, USA