UCL-Dehaze: Toward Real-World Image Dehazing via Unsupervised Contrastive Learning

Wang, Yongzhen; Yan, Xuefeng; Wang, Fu Lee; Xie, Haoran; Yang, Wenhan; Zhang, Xiao–Ping; Qin, Jing; Wei, Mingqiang

doi:10.1109/tip.2024.3362153

articleIEEE Transactions on Image ProcessingJan 1, 2024Closed access

UCL-Dehaze: Toward Real-World Image Dehazing via Unsupervised Contrastive Learning

YWYongzhen Wang XYXuefeng Yan FLFu Lee Wang HXHaoran Xie WYWenhan Yang

Nanjing University of Aeronautics and Astronautics · Anhui University of Technology · +4 more institutions

PubMed

Indexed incrossrefpubmed

Abstract

While the wisdom of training an image dehazing model on synthetic hazy data can alleviate the difficulty of collecting real-world hazy/clean image pairs, it brings the well-known domain shift problem. From a different yet new perspective, this paper explores contrastive learning with an adversarial training effort to leverage unpaired real-world hazy and clean images, thus alleviating the domain shift problem and enhancing the network's generalization ability in real-world scenarios. We propose an effective unsupervised contrastive learning paradigm for image dehazing, dubbed UCL-Dehaze. Unpaired real-world clean and hazy images are easily captured, and will serve as the important positive and negative samples…

Citation impact

167

total citations

FWCI: 36.88
Percentile: 100%
References: 71

Citations per year

Authors

8

Topics & keywords

Topics

Keywords

Computer science
Artificial intelligence
Image editing
Embedding
Image (mathematics)
Generalization
Leverage (statistics)
Deep learning

No related works found for this paper.