用CLIP指导去雾，仅微调部分层就能适配复杂雾霾场景。
Bilevel Layer-Positioning LoRA for Real Image Dehazing
- 通过CLIP跨模态对齐，无参考图也能实现无监督去雾。
- 自动搜索最优注入层，只更新关键网络层，节省计算资源。
- 在多个真实雾霾数据集上超越当前最佳方法，适合实际应用。
基于学习的图像去雾方法虽取得显著进展，但在多样真实雾霾场景中仍面临适应性挑战，主要源于缺乏有效的无监督机制和全模型微调带来的高成本。为此，我们提出一种由雾霾到清晰图像引导的文本导向损失，利用CLIP的跨模态能力将真实图像去雾重构为潜在空间中的语义对齐问题，从而在无参考图像情况下提供明确的无监督跨模态指导。此外，我们引入双层层位置定位低秩适配（BiLaLoRA）策略，同时学习LoRA参数并自动搜索最优注入层，实现对关键网络层的精准适配。大量实验表明，该方法在多个真实世界去雾基准测试中优于现有最先进方法。代码已公开于https://github.com/YanZhang-zy/BiLaLoRA。
原文摘要 · Abstract (English)
Learning-based real image dehazing methods have achieved notable progress, yet they still face adaptation challenges in diverse real haze scenes. These challenges mainly stem from the lack of effective unsupervised mechanisms for unlabeled data and the heavy cost of full model fine-tuning. To address these challenges, we propose the haze-to-clear text-directed loss that leverages CLIP's cross-modal capabilities to reformulate real image dehazing as a semantic alignment problem in latent space, thereby providing explicit unsupervised cross-modal guidance in the absence of reference images. Furthermore, we introduce the Bilevel Layer-positioning LoRA (BiLaLoRA) strategy, which learns both the LoRA parameters and automatically search the injection layers, enabling targeted adaptation of critical network layers. Extensive experiments demonstrate our superiority against state-of-the-art methods on multiple real-world dehazing benchmarks. The code is publicly available at https://github.com/YanZhang-zy/BiLaLoRA.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。