Boosting Flow-based Generative Super-Resolution Models via Learned Prior

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Tsao, Li-Yuan, Lo, Yi-Chen, Chang, Chia-Che, Chen, Hao-Wei, Tseng, Roy, Feng, Chien, Lee, Chun-Yi
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911891548274688
author Tsao, Li-Yuan
Lo, Yi-Chen
Chang, Chia-Che
Chen, Hao-Wei
Tseng, Roy
Feng, Chien
Lee, Chun-Yi
author_facet Tsao, Li-Yuan
Lo, Yi-Chen
Chang, Chia-Che
Chen, Hao-Wei
Tseng, Roy
Feng, Chien
Lee, Chun-Yi
contents Flow-based super-resolution (SR) models have demonstrated astonishing capabilities in generating high-quality images. However, these methods encounter several challenges during image generation, such as grid artifacts, exploding inverses, and suboptimal results due to a fixed sampling temperature. To overcome these issues, this work introduces a conditional learned prior to the inference phase of a flow-based SR model. This prior is a latent code predicted by our proposed latent module conditioned on the low-resolution image, which is then transformed by the flow model into an SR image. Our framework is designed to seamlessly integrate with any contemporary flow-based SR model without modifying its architecture or pre-trained weights. We evaluate the effectiveness of our proposed framework through extensive experiments and ablation analyses. The proposed framework successfully addresses all the inherent issues in flow-based SR models and enhances their performance in various SR scenarios. Our code is available at: https://github.com/liyuantsao/BFSR
format Preprint
id arxiv_https___arxiv_org_abs_2403_10988
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Boosting Flow-based Generative Super-Resolution Models via Learned Prior
Tsao, Li-Yuan
Lo, Yi-Chen
Chang, Chia-Che
Chen, Hao-Wei
Tseng, Roy
Feng, Chien
Lee, Chun-Yi
Computer Vision and Pattern Recognition
Artificial Intelligence
Flow-based super-resolution (SR) models have demonstrated astonishing capabilities in generating high-quality images. However, these methods encounter several challenges during image generation, such as grid artifacts, exploding inverses, and suboptimal results due to a fixed sampling temperature. To overcome these issues, this work introduces a conditional learned prior to the inference phase of a flow-based SR model. This prior is a latent code predicted by our proposed latent module conditioned on the low-resolution image, which is then transformed by the flow model into an SR image. Our framework is designed to seamlessly integrate with any contemporary flow-based SR model without modifying its architecture or pre-trained weights. We evaluate the effectiveness of our proposed framework through extensive experiments and ablation analyses. The proposed framework successfully addresses all the inherent issues in flow-based SR models and enhances their performance in various SR scenarios. Our code is available at: https://github.com/liyuantsao/BFSR
title Boosting Flow-based Generative Super-Resolution Models via Learned Prior
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2403.10988