TY - GEN
T1 - Could Human Gaze Augment Detectors of Synthetic Images?
AU - Fotopoulos, Nikolaos
AU - Riedmiller, Clara
AU - Bozkir, Efe
AU - Tsinganos, Panagiotis
AU - Ampeliotis, Dimitris
AU - Kasneci, Gjergji
AU - Kasneci, Enkelejda
AU - Skodras, Athanassios
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - Recent advances in generative adversarial networks (GANs) allow for the synthesis of extremely photo-realistic face images, deceiving even the most experienced observers, let alone the unsuspecting internet user. Due to this, there has been a considerable effort by the image forensics community to design appropriate tools for the detection of these images. This paper first implements one such detection technique based on spatial and cross-band co-occurrence matrices and convolutional neural networks (CNNs), and then attempts to improve it by introducing additional information obtained from the human gaze. We show that in cases where human observers correctly decide whether an image is real or fake, eye movement information in combination with spatial and cross-band co-occurrence matrices derived from observation regions can be informative towards the task of detecting fake images. However, only a limited increase in the detection accuracy is achieved.
AB - Recent advances in generative adversarial networks (GANs) allow for the synthesis of extremely photo-realistic face images, deceiving even the most experienced observers, let alone the unsuspecting internet user. Due to this, there has been a considerable effort by the image forensics community to design appropriate tools for the detection of these images. This paper first implements one such detection technique based on spatial and cross-band co-occurrence matrices and convolutional neural networks (CNNs), and then attempts to improve it by introducing additional information obtained from the human gaze. We show that in cases where human observers correctly decide whether an image is real or fake, eye movement information in combination with spatial and cross-band co-occurrence matrices derived from observation regions can be informative towards the task of detecting fake images. However, only a limited increase in the detection accuracy is achieved.
KW - Generative Adversarial Networks
KW - co-occurrence matrix
KW - eye tracking
KW - gaze
KW - image forensics
KW - synthetic / fake images
UR - http://www.scopus.com/inward/record.url?scp=85165501185&partnerID=8YFLogxK
U2 - 10.1109/DSP58604.2023.10167876
DO - 10.1109/DSP58604.2023.10167876
M3 - Conference contribution
AN - SCOPUS:85165501185
T3 - International Conference on Digital Signal Processing, DSP
BT - 2023 24th International Conference on Digital Signal Processing, DSP 2023
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 24th International Conference on Digital Signal Processing, DSP 2023
Y2 - 11 June 2023 through 13 June 2023
ER -