Deepfake Detection System Using Hybrid CNN–VIT Architecture
Article Sidebar
Main Article Content
The increasing use of deepfake media is a serious threat because highly realistic fake images and videos can be generated and posted online, leading to scams, identity theft, and misinformation. Because of how realistic these images and videos are, it is becoming increasingly difficult to distinguish them manually. Our project provides a softwarse application that can help determine whether an uploaded image or video is real or fake. It checks facial characteristics and the overall facial structure to identify characteristic differences between real and fake media. Additionally, it performs frequency analysis to identify characteristic artefacts that are usually created when digital processing of images occurs. To ensure that the results are interpretable, the application provides visual feedback about the facial areas that contributed to the determination, providing users with a clear understanding of why a particular image or video is real or fake. The project aims to develop a useful and easy to-use application that can be used for cybersecurity, digital forensics, and online media verification
Downloads
References
T. Nguyen and L. Tran, “Deepfake detection using CNN-based spatial and temporal features,” IEEE Access, vol. 12, pp. 3456–3472, 2024.
Y. Zhang and H. Chen, “Multi-model ensemble learning for robust deepfake detection,” IEEE Transactions on Artificial Intelligence, vol. 5, no. 2, pp. 112–130, 2024.
R. Kumar and P. Singh, “The arms race between deepfake generation and detection,” IEEE Signal Processing Magazine, vol. 41, no. 1, pp. 45–61, 2024.
J. Li and F. Zhao, “Advanced Deepfake Video Detection Using Temporal
Models,” in Proc. IEEE/CVF CVPR Workshops, pp. 78–92, 2024.
A. Rössler, D. Cozzolino, L. Verdoliva, C. Riess, J. Thies, and M. Nießner, “FaceForensics++: Learning to detect manipulated facial images,” in Proc. IEEE Int. Conf. Computer Vision (ICCV), pp. 1–11, 2019.
B. Dolhansky, R. Howes, B. Pflaum, N. Baram, and C. Ferrer, “The DeepFake detection challenge (DFDC) dataset,” in Proc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR), pp. 10438–10447, 2020.
D. Afchar, V. Nozick, J. Yamagishi, and I. Echizen, “MesoNet: A compact facial video forgery detection network,” in Proc. IEEE Int. Workshop on Information Forensics and Security (WIFS), pp. 1–7, 2018.
Y. He et al., “TransForensics: Image forgery localization with dense self-attention,” in Proc. IEEE/CVF Int. Conf. Computer Vision (ICCV), pp. 1501–1510, 2021.
Y. Liu, Y. Li, and S. Wang, “Vision transformer-based deepfake detection with frequency-aware learning,” IEEE Access, vol. 11, pp. 99872–99884, 2023.
C. Zhao, G. Wang, and X. Li, “Hybrid CNN–transformer framework for robust deepfake image detection,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 33, no. 9, pp. 4562–4575, 2023.
Y. K. Mali, V. U. Rathod, S. Dargad, V. N. Kapure, N. Vyawahare, and S. Gajarlewar, “Development of ROS (robotic operating system) for fault detection of underwater cables,” in Proc. 2023 IEEE 11th Region 10 Humanitarian Technology Conference (R10-HTC), pp. 956–961, 2023.
V. J. Shimpi and R. R. Ade, “Understanding 2016 drought in India by social media data mining,” Communications on Applied Electronics (CAE), vol. 5, no. 6, pp. 1–5, 2016.
A. Patle, V. Kapure, K. B. Chaure, A. S. Patil, V. R. Pujari, and H. P. Deshmukh, “Minefield evacuation proximity drift safety alert,” in Proc. 2024 International Conference on Computing, Sciences and Communications, 2024.
V. J. Shimpi and R. Ade, “A review: Social media data mining for understanding and analyzing different issues in society,” International Journal of Computer Applications.
This work is licensed under a Creative Commons Attribution 4.0 International License.
All articles published in our journal are licensed under CC-BY 4.0, which permits authors to retain copyright of their work. This license allows for unrestricted use, sharing, and reproduction of the articles, provided that proper credit is given to the original authors and the source.