The Unreasonable Effectiveness of Texture Transfer for Single Image Super-resolution

Muhammad Waleed Gondal, Bernhard Scholkopf, Michael Hirsch; Proceedings of the European Conference on Computer Vision (ECCV) Workshops, 2018, pp. 0-0

Abstract


While implicit generative models such as GANs have shown impressive results in high quality image reconstruction and manipulation using a combination of various losses, we consider a simpler approach leading to surprisingly strong results. We show that texture loss [1] alone allows the generation of perceptually high quality images. We provide a better understanding of texture constraining mechanism and develop a novel semantically guided texture constraining method for further improvement. Using a recently developed perceptual metric employing "deep features" and termed LPIPS [2], the method obtains state-of-theart results. Moreover, we show that a texture representation of those deep features better capture the perceptual quality of an image than the original deep features. Using texture information, off-the-shelf deep classification networks (without training) perform as well as the best performing (tuned and calibrated) LPIPS metrics.

Related Material


[pdf]
[bibtex]
@InProceedings{Gondal_2018_ECCV_Workshops,
author = {Waleed Gondal, Muhammad and Scholkopf, Bernhard and Hirsch, Michael},
title = {The Unreasonable Effectiveness of Texture Transfer for Single Image Super-resolution},
booktitle = {Proceedings of the European Conference on Computer Vision (ECCV) Workshops},
month = {September},
year = {2018}
}