Jul 2026
Backbone-Agnostic Stochastic Perturbation Learning for End-to-End Real-World Image Dehazing
Backbone-Agnostic Stochastic Perturbation Learning (BSPL) is proposed, a plug-and-play framework for end-to-end real-world image dehazing and a Dual-space Domain-diversified Distribution-aware Contrastive Loss to regularize both clean restoration and hazy reconstruction spaces with real-world and synthetic negatives.
Bingcai Wei, Yuning Cui, Mingyu Liu et al.
· arXiv.org · 0 citations