Showing 1–1 of 1 results for author: Chira, D
-
Image Super-Resolution With Deep Variational Autoencoders
Authors:
Darius Chira,
Ilian Haralampiev,
Ole Winther,
Andrea Dittadi,
Valentin Liévin
Abstract:
Image super-resolution (SR) techniques are used to generate a high-resolution image from a low-resolution image. Until now, deep generative models such as autoregressive models and Generative Adversarial Networks (GANs) have proven to be effective at modelling high-resolution images. VAE-based models have often been criticised for their feeble generative performance, but with new advancements such…
▽ More
Image super-resolution (SR) techniques are used to generate a high-resolution image from a low-resolution image. Until now, deep generative models such as autoregressive models and Generative Adversarial Networks (GANs) have proven to be effective at modelling high-resolution images. VAE-based models have often been criticised for their feeble generative performance, but with new advancements such as VDVAE, there is now strong evidence that deep VAEs have the potential to outperform current state-of-the-art models for high-resolution image generation. In this paper, we introduce VDVAE-SR, a new model that aims to exploit the most recent deep VAE methodologies to improve upon the results of similar models. VDVAE-SR tackles image super-resolution using transfer learning on pretrained VDVAEs. The presented model is competitive with other state-of-the-art models, having comparable results on image quality metrics.
△ Less
Submitted 26 October, 2022; v1 submitted 17 March, 2022;
originally announced March 2022.
