Flexible Style Image Super-Resolution using Conditional Objective

TL;DR

This paper introduces a flexible super-resolution method that uses a single CNN trained with a conditional objective to generate diverse high-resolution images based on style controls, avoiding multiple models.

Contribution

It proposes a novel training scheme with a conditional loss and a specialized architecture to produce multiple SR outputs from one model, enhancing efficiency and versatility.

Findings

Produces diverse SR outputs without artifacts

Achieves comparable quantitative performance to state-of-the-art methods

Enables local style-based control over super-resolution results

Abstract

Recent studies have significantly enhanced the performance of single-image super-resolution (SR) using convolutional neural networks (CNNs). While there can be many high-resolution (HR) solutions for a given input, most existing CNN-based methods do not explore alternative solutions during the inference. A typical approach to obtaining alternative SR results is to train multiple SR models with different loss weightings and exploit the combination of these models. Instead of using multiple models, we present a more efficient method to train a single adjustable SR model on various combinations of losses by taking advantage of multi-task learning. Specifically, we optimize an SR model with a conditional objective during training, where the objective is a weighted sum of multiple perceptual losses at different feature levels. The weights vary according to given conditions, and the set of weights is defined as a style controller. Also, we present an architecture appropriate for this training scheme, which is the Residual-in-Residual Dense Block equipped with spatial feature transformation layers. At the inference phase, our trained model can generate locally different outputs conditioned on the style control map. Extensive experiments show that the proposed SR model produces various desirable reconstructions without artifacts and yields comparable quantitative performance to state-of-the-art SR methods.