Papers › Recovering Realistic Texture in Image Super-resolution by Deep Spatial Feature Transform

Recovering Realistic Texture in Image Super-resolution by Deep Spatial Feature Transform

9 Apr 2018CVPR 2018 6arXiv:1804.02815archive 2025-07-28

Xintao Wang, Ke Yu, Chao Dong, Chen Change Loy

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 and EnhanceNet.

Paper PDF Conference PDF Code

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

xinntao/SFTGAN officialmentioned on GitHubpytorch report

micmic123/qmapcompression mentioned on GitHubpytorch report

sdauzcm/sr-basicsr mentioned on GitHubpytorchApache-2.0 report

xinntao/BasicSR mentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Image Super-Resolution Semantic Segmentation Super-Resolution

Datasets

Introduced by this paper, per the archive.

OST300

Results from the paper archive 2025-07-28

Task	Dataset	Model	Metric	Value	Rank at snapshot	Leaderboard	Report
Image Super-Resolution	BSD100 - 4x upscaling	SFT-GAN	PSNR	25.33	#62 of 71	Archive leaderboard	report
Image Super-Resolution	BSD100 - 4x upscaling	SFT-GAN	SSIM	0.651	#62 of 71	Archive leaderboard	report
Image Super-Resolution	Set14 - 4x upscaling	SFT-GAN	PSNR	26.13	#99 of 104	Archive leaderboard	report
Image Super-Resolution	Set14 - 4x upscaling	SFT-GAN	SSIM	0.694	#99 of 104	Archive leaderboard	report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Introduced by this paper: Spatial Feature Transform

Batch Normalization Convolution Dense Connections Dropout Max Pooling PReLU PixelShuffle ReLU Residual Block Residual Connection SPADE SRGAN SRGAN Residual Block Sigmoid Activation Softmax Spatial Feature Transform VGG Loss

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections