IID-Net: Image Inpainting Detection Network via Neural Architecture Search and Attention

Abstract

Deep learning (DL) has demonstrated its powerful capabilities in the field of image inpainting, which could produce visually plausible results. Meanwhile, the malicious use of advanced image inpainting tools (e.g. removing key objects to report fake news, erasing visible copyright watermarks, etc.) has led to increasing threats to the reliability of image data. To fight against the inpainting forgeries (not only DL-based but also traditional ones), in this work, we propose a novel end-to-end Image Inpainting Detection Network (IID-Net), to detect the inpainted regions at pixel accuracy. The proposed IID-Net consists of three sub-blocks: the enhancement block, the extraction block and the decision block. Specifically, the enhancement block aims to enhance the inpainting traces by using hierarchically combined special layers. The extraction block, automatically designed by Neural Architecture Search (NAS) algorithm, is targeted to extract features for the actual inpainting detection tasks. To further optimize the extracted latent features, we integrate global and local attention modules in the decision block, where the global attention reduces the intra-class differences by measuring the similarity of global features, while the local attention strengthens the consistency of local features. Furthermore, we thoroughly study the generalizability of our IID-Net, and find that different training data could result in vastly different generalization capability. By carefully examining 10 popular inpainting methods, we identify that the IID-Net trained on only one specific deep inpainting method exhibits desirable generalizability; namely, the obtained IID-Net can accurately detect and localize inpainting manipulations for various unseen inpainting methods as well. Extensive experimental results are presented to validate the superiority of the proposed IID-Net, compared with the state-of-the-art competitors. Our results would suggest that common artifacts are shared across diverse image inpainting methods. Finally, we build a public inpainting dataset of 10K image pairs for future research in this area.