Image Forgery Localization Based on Multi-Scale Convolutional Neural Networks

In this paper, we propose to use Multi-Scale Convolutional Neural Networks (CNNs) to conduct forgery localization in digital image forensics. A unified CNN architecture for input sliding windows of different scales is designed. Then, we elaborately design the training procedures of CNNs on sampled training patches in the IEEE IFS-TC Image Forensics Challenge training images. With a set of carefully trained multi-scale CNNs, tampering possibility maps generated by these CNNs are fused to get the final decision maps. By exploiting the benefits of both the small-scale and large-scale analyses, the multi-scale analysis can lead to a performance leap for forgery localization of CNNs. Numerous experiments are conducted to demonstrate the effectiveness and efficiency of the proposed method.