Deep Back-Projection Networks For Super-Resolution

TL;DR

Deep Back-Projection Networks (DBPN) introduce iterative up- and down-sampling with error feedback for improved super-resolution, achieving state-of-the-art results especially at large scaling factors like 8x.

Contribution

The paper presents a novel back-projection architecture with iterative error correction and dense feature concatenation for enhanced super-resolution performance.

Findings

Achieves superior super-resolution results at large scaling factors.

Establishes new state-of-the-art performance for 8x super-resolution.

Demonstrates effectiveness across multiple datasets.

Abstract

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.