Student : K Crawford
About the student
I’m a Computer Science Honours student passionate about building systems that tackle real-world problems through cutting-edge solutions.
About the Project
This project focuses on detecting whether an image has been edited or tampered with. It combines three different approaches to make the results more reliable. The first is a U-Net neural network, which analyzes the image and produces a heatmap that highlights areas likely to have been edited. The second is Error Level Analysis (ELA), which works by re-saving the image at different compression levels where edited regions often compress differently from the rest of the picture, exposing hidden inconsistencies. The third method is metadata analysis, which examines the hidden information stored in image. This can reveal whether editing software was used, whether timestamps don’t match, or even if the image contains AI-generation tags. By combining all three techniques, the system not only detects whether an image has been tampered with but also shows where the changes are.
