Copyright © 2026 Authors retain the copyright of this article. This article 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.

@article{192072,
        author = {Shivaprasad Satla},
        title = {A Deep Reconstruction-Based Scalable Framework for Image Forgery Detection},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {9},
        pages = {121-129},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=192072},
        abstract = {Currently, manipulation of digital images with modern powerful image editing software is a commonplace approach. Therefore, determining the authenticity of the image is one of the crucial challenges faced by multimedia forensics. The presented project incorporates RIFD-Net, a deep-learning-based system for detecting and localizing forged regions in images. At the heart of the system is an encoder-decoder network, inspired by U-Net architecture, performing pixel-level analysis to generate forgery masks visually highlighting tampered areas. In addition to the deep learning-based detection of forgeries, this system provides a REST API to enable real-time forgery confidence estimation that allows seamless connection with external applications and large-scale image databases. An EXIF metadata analysis module has also been included, which could identify missing or suspicious metadata patterns of images, hinting at tampering. Perceptual hashing for efficient detection of duplicate and near-duplicate images is conducted. The proposed model is first trained on paired original images along with ground-truth masks and optimized by using the Adam optimizer with mean squared error loss. A web interface using Streamlit will be presented that will enable real-time image uploading, analysis, and visualization. Experimental results establish that the system will be able to combine the complementary visual, metadata, and structural cues through effective processing, making it practical and scalable for modern digital image forensics.},
        keywords = {Deep Learning, EXIF Analysis, Forgery Mask, Image Forgery, Image Splicing, Noise Detection, Perceptual Hashing, U-Net Architecture},
        month = {January},
        }