DIGITAL IMAGE FORGERY DETECTION USING DEEP LEARNING

Abstract

Digital image forgery poses a significant threat to the integrity of visual content, 
necessitating robust and efficient forgery detection mechanisms. This project 
introduces an innovative approach to image forgery detection through the fusion of 
lightweight deep learning models. Leveraging architectures like SqueezeNet, 
MobileNetV2, and ShuffleNet, the proposed system achieves a delicate balance 
between accuracy and computational efficiency. The fusion methodology enhances 
the 
system's resilience against a variety of forgery techniques, ensuring 
comprehensive analysis of diverse image features. Experimental results demonstrate 
the system's efficacy in identifying manipulated images, making it suitable for real
time applications. This project not only contributes to the evolving landscape of 
multimedia forensics but also provides a resource-efficient solution for combating the 
rising threat of digital image manipulation.