Accelerating Image Super-Resolution Networks with Pixel-Level Classification

TL;DR

This paper introduces PCSR, a pixel-level classification method for super-resolution that adaptively allocates computational resources to individual pixels, improving efficiency and performance without retraining.

Contribution

The paper presents a novel pixel-level classifier that dynamically assigns upsamplers to pixels based on difficulty, enhancing super-resolution efficiency and accuracy over patch-based methods.

Findings

PCSR outperforms existing patch-distributing methods in PSNR-FLOP trade-offs.

It maintains performance while reducing computational costs.

The approach is effective across different backbone models and benchmarks.

Abstract

In recent times, the need for effective super-resolution (SR) techniques has surged, especially for large-scale images ranging 2K to 8K resolutions. For DNN-based SISR, decomposing images into overlapping patches is typically necessary due to computational constraints. In such patch-decomposing scheme, one can allocate computational resources differently based on each patch's difficulty to further improve efficiency while maintaining SR performance. However, this approach has a limitation: computational resources is uniformly allocated within a patch, leading to lower efficiency when the patch contain pixels with varying levels of restoration difficulty. To address the issue, we propose the Pixel-level Classifier for Single Image Super-Resolution (PCSR), a novel method designed to distribute computational resources adaptively at the pixel level. A PCSR model comprises a backbone, a pixel-level classifier, and a set of pixel-level upsamplers with varying capacities. The pixel-level classifier assigns each pixel to an appropriate upsampler based on its restoration difficulty, thereby optimizing computational resource usage. Our method allows for performance and computational cost balance during inference without re-training. Our experiments demonstrate PCSR's advantage over existing patch-distributing methods in PSNR-FLOP trade-offs across different backbone models and benchmarks. The code is available at https://github.com/3587jjh/PCSR.