Meta-SR: A Magnification-Arbitrary Network for Super-Resolution

TL;DR

Meta-SR introduces a novel neural network that enables super-resolution at any scale factor, including non-integer values, using a single model that dynamically predicts upscale filter weights.

Contribution

The paper proposes Meta-SR, the first model capable of arbitrary scale super-resolution with a dynamic Meta-Upscale Module that predicts filter weights based on scale factors.

Findings

Outperforms existing methods on benchmark datasets

Supports continuous zoom-in with a single model

Effective for both integer and non-integer scale factors

Abstract

Recent research on super-resolution has achieved great success due to the development of deep convolutional neural networks (DCNNs). However, super-resolution of arbitrary scale factor has been ignored for a long time. Most previous researchers regard super-resolution of different scale factors as independent tasks. They train a specific model for each scale factor which is inefficient in computing, and prior work only take the super-resolution of several integer scale factors into consideration. In this work, we propose a novel method called Meta-SR to firstly solve super-resolution of arbitrary scale factor (including non-integer scale factors) with a single model. In our Meta-SR, the Meta-Upscale Module is proposed to replace the traditional upscale module. For arbitrary scale factor, the Meta-Upscale Module dynamically predicts the weights of the upscale filters by taking the scale factor as input and use these weights to generate the HR image of arbitrary size. For any low-resolution image, our Meta-SR can continuously zoom in it with arbitrary scale factor by only using a single model. We evaluated the proposed method through extensive experiments on widely used benchmark datasets on single image super-resolution. The experimental results show the superiority of our Meta-Upscale.