Vision-Language Gradient Descent-driven All-in-One Deep Unfolding Networks

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zeng, Haijin, Wang, Xiangming, Chen, Yongyong, Su, Jingyong, Liu, Jie
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910888393441280
author Zeng, Haijin
Wang, Xiangming
Chen, Yongyong
Su, Jingyong
Liu, Jie
author_facet Zeng, Haijin
Wang, Xiangming
Chen, Yongyong
Su, Jingyong
Liu, Jie
contents Dynamic image degradations, including noise, blur and lighting inconsistencies, pose significant challenges in image restoration, often due to sensor limitations or adverse environmental conditions. Existing Deep Unfolding Networks (DUNs) offer stable restoration performance but require manual selection of degradation matrices for each degradation type, limiting their adaptability across diverse scenarios. To address this issue, we propose the Vision-Language-guided Unfolding Network (VLU-Net), a unified DUN framework for handling multiple degradation types simultaneously. VLU-Net leverages a Vision-Language Model (VLM) refined on degraded image-text pairs to align image features with degradation descriptions, selecting the appropriate transform for target degradation. By integrating an automatic VLM-based gradient estimation strategy into the Proximal Gradient Descent (PGD) algorithm, VLU-Net effectively tackles complex multi-degradation restoration tasks while maintaining interpretability. Furthermore, we design a hierarchical feature unfolding structure to enhance VLU-Net framework, efficiently synthesizing degradation patterns across various levels. VLU-Net is the first all-in-one DUN framework and outperforms current leading one-by-one and all-in-one end-to-end methods by 3.74 dB on the SOTS dehazing dataset and 1.70 dB on the Rain100L deraining dataset.
format Preprint
id arxiv_https___arxiv_org_abs_2503_16930
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Vision-Language Gradient Descent-driven All-in-One Deep Unfolding Networks
Zeng, Haijin
Wang, Xiangming
Chen, Yongyong
Su, Jingyong
Liu, Jie
Computer Vision and Pattern Recognition
Dynamic image degradations, including noise, blur and lighting inconsistencies, pose significant challenges in image restoration, often due to sensor limitations or adverse environmental conditions. Existing Deep Unfolding Networks (DUNs) offer stable restoration performance but require manual selection of degradation matrices for each degradation type, limiting their adaptability across diverse scenarios. To address this issue, we propose the Vision-Language-guided Unfolding Network (VLU-Net), a unified DUN framework for handling multiple degradation types simultaneously. VLU-Net leverages a Vision-Language Model (VLM) refined on degraded image-text pairs to align image features with degradation descriptions, selecting the appropriate transform for target degradation. By integrating an automatic VLM-based gradient estimation strategy into the Proximal Gradient Descent (PGD) algorithm, VLU-Net effectively tackles complex multi-degradation restoration tasks while maintaining interpretability. Furthermore, we design a hierarchical feature unfolding structure to enhance VLU-Net framework, efficiently synthesizing degradation patterns across various levels. VLU-Net is the first all-in-one DUN framework and outperforms current leading one-by-one and all-in-one end-to-end methods by 3.74 dB on the SOTS dehazing dataset and 1.70 dB on the Rain100L deraining dataset.
title Vision-Language Gradient Descent-driven All-in-One Deep Unfolding Networks
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2503.16930