High-Resolution Image Inpainting using Multi-Scale Neural Patch Synthesis
This addresses the challenge of generating high-quality, detailed inpaintings for high-resolution images, which is crucial for applications like object removal in image editing, though it appears incremental as it builds on prior deep learning techniques.
The paper tackles the problem of high-resolution image inpainting, where existing methods produce blurry results due to memory and training limitations, and proposes a multi-scale neural patch synthesis approach that achieves state-of-the-art accuracy on datasets like ImageNet and Paris Streetview, producing sharper and more coherent inpainted regions.
Recent advances in deep learning have shown exciting promise in filling large holes in natural images with semantically plausible and context aware details, impacting fundamental image manipulation tasks such as object removal. While these learning-based methods are significantly more effective in capturing high-level features than prior techniques, they can only handle very low-resolution inputs due to memory limitations and difficulty in training. Even for slightly larger images, the inpainted regions would appear blurry and unpleasant boundaries become visible. We propose a multi-scale neural patch synthesis approach based on joint optimization of image content and texture constraints, which not only preserves contextual structures but also produces high-frequency details by matching and adapting patches with the most similar mid-layer feature correlations of a deep classification network. We evaluate our method on the ImageNet and Paris Streetview datasets and achieved state-of-the-art inpainting accuracy. We show our approach produces sharper and more coherent results than prior methods, especially for high-resolution images.