A Review on Digital Image Forgery Detection Using Deep Learning, Transfer Learning, and Hybrid Techniques

Abstract

Digital images have become an essential part of everyday life, appearing in social media, news, medical, and legal contexts. However, as image-editing tools have advanced, it has become increasingly easy to alter or fabricate images. These manipulations, such as copy-move, splicing, and retouching, can spread misinformation or serve as false evidence, making image forgery detection a critical area of research. Traditional detection methods relied on manually crafted features like noise patterns or color inconsistencies, which often proved ineffective when images were compressed, resized, or modified in complex ways.In recent years, deep learning has revolutionized this field by enabling models to automatically learn useful features from data. Transfer learning, in particular, leverages pre-trained con volutional neural networks such as VGG, ResNet, and MobileNet to achieve higher accuracy, even with smaller datasets.This review examines the increasing use of transfer learning in digital image forgery detection.It discusses how techniques like Error Level Analysis (ELA) and recompression-based preprocessing help identify forgery clues, while Grad-CAM visualization aids in interpreting model decisions by highlighting manipulated regions. The paper concludes by emphasizing the need for future systems that balance accuracy, interpretability, and efficiency to provide reliable and explainable solutions for real-world image forgery detection.

Introduction

The text examines the growing challenge of digital image forgery in an era where advanced editing tools and AI-based image generation make image manipulation easy and difficult to detect. Since images are widely trusted as evidence in media, politics, and law, forged images pose serious risks by spreading misinformation, damaging reputations, and influencing public opinion. Common forgeries such as copy-move and splicing are often seamless and hard to identify through human inspection.

Early forgery detection methods relied on handcrafted features like color inconsistencies, noise patterns, and compression artifacts, but these approaches struggled with complex manipulations and varying image conditions. The adoption of deep learning, particularly Convolutional Neural Networks (CNNs), has significantly improved detection performance by automatically learning complex spatial and contextual features. Transfer learning using pre-trained models such as VGG, ResNet, DenseNet, and MobileNet has further enhanced accuracy, efficiency, and generalization, especially when training data is limited.

The literature survey highlights a wide range of modern approaches, including CNN-based classifiers, hybrid models combining handcrafted and deep features, multi-domain and multi-stream architectures, and lightweight models suitable for real-time or resource-constrained environments. Preprocessing techniques like Error Level Analysis and recompression amplify forgery traces, while explainable AI tools such as Grad-CAM improve transparency by visualizing detected tampered regions. Recent methods also extend to advanced tasks such as deepfake detection, document image forensics, and privacy-preserving scenarios.

Overall, the analysis shows a clear progression from traditional feature-based methods to robust deep learning and transfer learning frameworks. Hybrid architectures, feature fusion, and explainable AI have led to high accuracy and improved reliability. Future research is directed toward enhancing interpretability, reducing computational cost, and ensuring scalability across diverse image domains to enable trustworthy, real-time image forgery detection systems.

Conclusion

This paper reviewed several transfer-learning-based approaches for digital image forgery detection. Studies demonstrate that ELA preprocessing combined with lightweight CNN backbones like MobileNetV2 improves detection accuracy [2], [5]. However, interpretability and multi-forgery capability remain open research problems [18]. Future research should emphasize explainable AI models using Grad-CAM [16] and hybrid CNN architectures [13], [14] to enhance trust and usability in real-world applications.

References

[1] Y. E. Abdalla, M. T. Iqbal, and M. Shehata, “Image forgery detection based on deep transfer learning,” European Journal of Electrical and Computer Engineering, vol. 3, no. 5, pp. 1–8, 2019. [2] A. Hebbar and A. Kunte, “Transfer learning approach for splicing and copy-move image tampering detection,” in Proc. IEEE International Conference on Electronics, Computing and Communication Technologies, 2021. [3] I.-J. Yu, S.-H. Nam, W. Ahn, M.-J. Kwon, and H.-K. Lee, “Manipulation classification for JPEG images using multi-domain features,” IEEE Access, vol. 8, pp. 210837–210854, 2020. [4] S. S. Ali, I. I. Ganapathi, N.-S. Vu, S. D. Ali, N. Saxena, and N. Werghi, “Image forgery detection using deep learning by recompressing images,” Electronics, vol. 11, no. 3, p. 403, 2022. [5] T. Khalil, S. A. Ali, and M. A. Khan, “Enhancing digital image forgery detection using transfer learning,” IEEE Access, vol. 11, pp. 91583 91594, 2023. [6] A. R. Al-Shamasneh and R. W. Ibrahim, “Image splicing forgery detection using feature-based of Sonine functions and deep features,” Journal of King Saud University-Computer and Information Sciences, vol. 36, no. 1, p. 101925, 2024.. [7] M. F. Abdulqader, A. Y. Dawod, and A. Z. Ablahd, “Detection of tamper forgery image in security digital image,” Measurement: Sensors, vol. 27, p. 100746, 2023. [8] H. B. Macit and A. Koyun, “An active image forgery detection approach based on edge detection,” in Proc. IEEE International Conference on Innovations in Intelligent Systems and Applications, 2023. [9] W. Li, B. Li, K. Zheng, S. Li, and H. Li, “Document image forgery detection and localization in desensitization scenarios,” Engineering Applications of Artificial Intelligence, vol. 128, p. 107541, 2025. [10] Z. Liang, W. Liu, R. Wang, M. Wu, B. Li, Y. Zhang, L. Wang, and X. Yang, “Transfer learning of real image features with soft contrastive loss for fake image detection,” Engineering Applications of Artificial Intelligence, vol. 133, p. 108542, 2024. [11] E. U. H. Qazi, T. Zia, and A. Almorjan, “Deep learning-based digital image forgery detection system,” International Journal of Intelligent Automation and Soft Computing, vol. 34, no. 2, pp. 1125–1140, 2022. [12] A. K. Jaiswal and S. Kumar, “An image forgery detection approach based on convolutional neural networks using transfer learning,” Journal of Intelligent Systems, vol. 32, no. 1, pp. 45–58, 2023. [13] Y. Zhu, X. Wang, and L. Chen, “Image tampering detection based on RDS-YOLOv5 feature enhancement transformation,” Scientific Reports, vol. 14, no. 1, p. 5678, 2024. [14] H. Liu, W. Zhang, and R. Yang, “Hierarchical progressive image forgery detection and localization method based on UNet (HPUNet),” Big Data and Cognitive Computing, vol. 8, no. 3, p. 234, 2024. [15] W. Ye, J. Li, and H. Wang, “A two-stage detection method of copy-move forgery based on parallel feature fusion,” EURASIP Journal on Wireless Communications and Networking, vol. 2022, no. 1, p. 45, 2022. [16] R. R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, and D. Batra, “Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization,” in Proc. IEEE International Conference on Computer Vision (ICCV), pp. 618–626, 2017. [17] M. Karthika, P. Saranya, R. Sudhakar, V. Uma, and R. Kiruthika, “Image Forgery Detection using VGG16-UNet and Error Level Analysis (ELA),” in Procedia Computer Science, vol. 232, pp. 1502–1510, 2024. [18] S. Agarwal, P. Singh, and D. S. Chauhan, “A Comprehensive Review of Deep Learning-Based Methods for Image Forensics,” in Journal of Imaging, vol. 7, no. 11, pp. 1–20, 2021. [19] R. Kaur, A. Kaur, and R. Vig, “Copy-Move Forgery Detection using Deep Learning for Image and Video Forensics,” in Journal of Imaging, vol. 7, no. 5, pp. 1–14, 2021.. [20] A. Khatoon, S. Khan, and A. Rehman, “The Effect of Error Level Analysis on Image Forgery Detection Using Deep Learning,” in Kinetik: Game Technology, Information System, Computer Network, Computing, Electronics, and Control, vol. 6, no. 2, pp. 195–202, 2021

Copyright

Copyright © 2025 Ajna Ashraf, Arsha K B, Navami P M, T S Rithuparna, Ms. Mahshiya Mishab. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.