Paper Info
Title
Recovering Realistic Texture in Image Super-Resolution by Deep Spatial Feature Transform
Abstract
Despite that convolutional neural networks (CNN) have recently demonstrated high-quality reconstruction for single-image super-resolution (SR), recovering natural and realistic texture remains a challenging problem. In this paper, we show that it is possible to recover textures faithful to semantic classes. In particular, we only need to modulate features of a few intermediate layers in a single network conditioned on semantic segmentation probability maps. This is made possible through a novel Spatial Feature Transform (SFT) layer that generates affine transformation parameters for spatial-wise feature modulation. SFT layers can be trained end-to-end together with the SR network using the same loss function. During testing, it accepts an input image of arbitrary size and generates a high-resolution image with just a single forward pass conditioned on the categorical priors. Our final results show that an SR network equipped with SFT can generate more realistic and visually pleasing textures in comparison to state-of-the-art SRGAN [27] and EnhanceNet [38].
Year
DOI
Venue
2018
10.1109/CVPR.2018.00070
2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition
Keywords
DocType
Volume
high-resolution image,SR network,deep spatial feature,convolutional neural networks,CNN,high-quality reconstruction,single-image super-resolution,semantic classes,semantic segmentation probability maps,transformation parameters,spatial-wise feature modulation,SFT layers,input image,spatial feature transform layer,affine transformation parameters,SRGAN,EnhanceNet,categorical priors
Conference
abs/1804.02815
ISSN
ISBN
Citations 
1063-6919
978-1-5386-6421-6
29
PageRank 
References 
Authors
0.87
31
4
Name
Order
Citations
PageRank
Xintao Wang11449.14
Ke Yu215218.85
Chao Dong3206480.72
Chen Change Loy44484178.56