Loading article...
Loading article...
Stream/Area:
-
Subject:
-
Paper ID:
IJSREM65065
Volume:
10 |
6 |
Jun 2026
Pages:
1-5
Publication Frequency:
Monthly (12 issues per year)
License:International Journal of Scientific Research in Engineering and Management follows the Creative Commons Attribution 4.0 (CC BY 4.0) License, allowing free use and distribution with proper attribution.
Cyber Forensic And Information Security
Dr. M.G.R. Educational And Research Institute
Assistant Professor , Cyber Forensic And Information Security
Dr. M.G.R. Educational And Research Institute
Assistant Professor ,Cyber Forensic and Information Security
Dr. M.G.R. Educational And Research Institute
The creation of images has become a lot easier with the development of Generative Adversarial Networks and diffusion-based models. These tools are great for creativity. They also pose a big threat to our trust in digital information, our personal privacy and the accuracy of the information we see on social media. It has become really important to detect these images in real time and this is a big challenge for computer vision and cybersecurity research. This paper is about a framework that uses deep learning to detect fake images. The framework is designed to be fast and accurate making it great for use in time. It uses a backbone to look at the basic features of an image and a multi-head self-attention mechanism that helps the network focus on the parts of the image that do not look right. These parts can be things like blending around the face unnatural skin textures and lighting that does not look normal. The model is trained on a dataset that includes lots of different types of fake images including FaceForensics++ Celeb-DF and other synthetic images. This helps the model to work with all kinds of fake images. The model also has an attention refinement module that helps to reduce the amount of computing power it needs without making it less accurate. This means it can be used on all kinds of devices those with limited power. The results of the experiments show that the framework is really accurate with a detection accuracy of 97.4% when tested on datasets. It is also faster than methods making it great for use in real time. The study also looked at how the model makes its predictions using heatmaps to show which parts of the image it is looking at. This work is a step in detecting fake images and it provides a practical solution, to this growing problem.
KEYWORDS
This paper is about a way to detect fake images in real time. It uses a kind of deep learning that combines the strengths of two different methods. The first method is called neural networks, which is really good at looking at the details of an image. The second method is called -head self-attention, which is good at understanding the big picture. The new system is really good at detecting images and it can do it very quickly. It can look at 38 images per second. The system also has a part that helps it find fake images that are really hard to spot. We can see how the system makes its decisions by looking at a kind of map that it creates. As people get better at making images and videos we need to have systems that can detect them. We need these systems to be able to explain how they make their decisions. This is important, for keeping the internet safe and honest. In the future we want to make this system even better. We want to be able to detect videos, not just images. We also want to find a way to use this system without having to share all of our data with each other. This will help keep our information private.
[1] K. Shiohara and T. Yamasaki, "Detecting Deepfakes with Self-Blended Images," in Proc. IEEE/CVF Conf. Computer Vision and Pattern Recognition (CVPR), 2022. [2] A. Khormali and J.-S. Yuan, "DFDT: An End-to-End DeepFake Detection Framework Using Vision Transformer," Applied Sciences, vol. 12, no. 6, 2022. [3] S. Ganguly, S. Mohiuddin, S. Malakar, E. Cuevas, and R. Sarkar, "Visual Attention-Based Deepfake Video Forgery Detection," Pattern Analysis and Applications, vol. 25, no. 4, pp. 981–992, 2022. [4] P. Prajapati and C. Pollett, "MRI-GAN: A Generalized Approach to Detect DeepFakes Using Perceptual Image Assessment," arXiv preprint arXiv:2203.00108, 2022. [5] S. Ge, F. Lin, C. Li, D. Zhang, W. Wang, and D. Zeng, "Deepfake Video Detection via Predictive Representation Learning," ACM Trans. Multimedia Computing, Communications, and Applications, vol. 18, no. 2s, pp. 1–21, 2022. [6] J. Zhang, K. Cheng, G. Sovernigo, and X. Lin, "A Heterogeneous Feature Ensemble Learning Based Deepfake Detection Method," in Proc. IEEE Int. Conf. Communications (ICC), 2022, pp. 2084–2089. [7] D. A. Coccomini, N. Messina, C. Gennaro, and F. Falchi, "Combining EfficientNet and Vision Transformers for Video Deepfake Detection," in Image Analysis and Processing – ICIAP 2022, Springer, 2022. [8] Z. Yang, J. Liang, Y. Xu, X.-Y. Zhang, and R. He, "Masked Relation Learning for DeepFake Detection," IEEE Trans. Information Forensics and Security, 2023. [9] H. Zhao, W. Zhou, D. Chen, T. Wei, W. Zhang, and N. Yu, "Multi-Attentional Deepfake Detection," in Proc. IEEE/CVF Conf. Computer Vision and Pattern Recognition (CVPR), 2023. [10] N. Sandotra and B. Arora, "A Comprehensive Evaluation of Feature-Based AI Techniques for Deepfake Detection," Neural Computing and Applications, 2023.
Publication Date
6/16/2026
Peer Review Type
Double-Blind Peer Review
Review Rounds: 2
Documents
Article Views
1
Total views since publication
PDF Downloads
3
Total downloads since publication
Geographic Reach
5
Digital Preservation: This article is permanently archived on Zenodo and complies with all data preservation standards across major indexing platforms.
Plagiarism Screening
This manuscript has been checked for plagiarism using TurnitinPlagiarismChecker software.
Conflict of Interest
The authors declare no conflict of interest related to this research work.
Funding Information
No funding information available for this research.
Related research in the same subject area
Loading similar articles...