Looking at the images in this paper, it’s easy to see a correlation between machine learning and archeology. The authors propose a new image problem: using small patches of an image to reconstruct the full image. They train a custom GAN with a spatial loss on multiple object-specific datasets including faces, waterfalls, cars, and ceramics and show its ability to generate images from the small patches. To fuel future research, the datasets that were used in this paper were also published.
Reassembling Image Fragments
Post · Apr 3, 2017 16:46 · Share on Twitter
We introduce a new problem of generating an image based on a small number of key local patches without any geometric prior. In this work, key local patches are defined as informative regions of the target object or scene. This is a challenging problem since it requires generating realistic images and predicting locations of parts at the same time. We construct adversarial networks to tackle this problem. A generator network generates a fake image as well as a mask based on the encoder-decoder framework. On the other hand, a discriminator network aims to detect fake images. The network is trained with three losses to consider spatial, appearance, and adversarial information. The spatial loss determines whether the locations of predicted parts are correct. Input patches are restored in the output image without much modification due to the appearance loss. The adversarial loss ensures output images are realistic. The proposed network is trained without supervisory signals since no labels of key parts are required. Experimental results on six datasets demonstrate that the proposed algorithm performs favorably on challenging objects and scenes.