A General Method to Incorporate Spatial Information into Loss Functions for GAN-based Super-resolution Models

arXiv:2403.10589v1 Announce Type: cross
Abstract: Generative Adversarial Networks (GANs) have shown great performance on super-resolution problems since they can generate more visually realistic images and video frames. However, these models often introduce side effects into the outputs, such as unexpected artifacts and noises. To reduce these artifacts and enhance the perceptual quality of the results, in this paper, we propose a general method that can be effectively used in most GAN-based super-resolution (SR) models by introducing essential spatial information into …

abstract adversarial arxiv cs.cv eess.iv effects functions gan gans general generate generative generative adversarial networks however images information loss networks performance reduce spatial type video