Deep Back-Projection Networks For Super-Resolution

The feed-forward architectures of recently proposed deep super-resolution networks learn representations of low-resolution inputs, and the non-linear mapping from those to high-resolution output. However, this approach does not fully address the mutual dependencies of low- and high-resolution images. We propose Deep Back-Projection Networks (DBPN), that exploit iterative up- and down-sampling layers, providing an error feedback mechanism for projection errors at each stage. We construct mutually-connected up- and down-sampling stages each of which represents different types of image degradation and high-resolution components. We show that extending this idea to allow concatenation of features across up- and down-sampling stages (Dense DBPN) allows us to reconstruct further improve super-resolution, yielding superior results and in particular establishing new state of the art results for large scaling factors such as 8x across multiple data sets.

PDF Abstract CVPR 2018 PDF CVPR 2018 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Image Super-Resolution BSD100 - 4x upscaling D-DBPN PSNR 27.72 # 18
SSIM 0.740 # 24
Image Super-Resolution Manga109 - 4x upscaling D-DBPN SSIM 0.914 # 20
Image Super-Resolution Set14 - 4x upscaling D-DBPN PSNR 28.82 # 26
SSIM 0.786 # 32
Image Super-Resolution Urban100 - 4x upscaling D-DBPN SSIM 0.795 # 24
Video Super-Resolution Vid4 - 4x upscaling DBPN PSNR 25.37 # 16
SSIM 0.737 # 16


No methods listed for this paper. Add relevant methods here