概念感知低秩适配用于域对齐的分割数据集生成
Concept-Aware LoRA for Domain-Aligned Segmentation Dataset Generation
CA-LoRA 通过选择性更新与目标域概念（如风格、视角）相关的权重，在微调 T2I 模型时保留预训练知识，从而生成多样且对齐的分割数据集；该思路可复用于其他需要域对齐数据生成的视觉任务，并提升域泛化性能。
原文依据 we propose Concept-Aware LoRA (CA-LoRA), a novel fine-tuning approach that selectively identifies and updates only the weights associated with necessary concepts
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论本文通过文本到图像（T2I）生成模型生成数据集，以降低图像采集和标注成本，从而解决语义分割中的数据稀缺问题。分割数据集生成面临两个关键挑战：1）使生成样本与目标域对齐；2）生成训练数据之外的信息性样本。微调 T2I 模型有助于生成与目标域对齐的样本，但往往过拟合和记忆训练数据，限制了其生成多样且良好对齐样本的能力。为克服这些问题，我们提出概念感知低秩适配（CA-LoRA），一种新颖的微调方法，它选择性地识别并仅更新与域对齐所需概念（如风格或视角）相关的权重，同时保留 T2I 模型的预训练知识以生成信息性样本。我们证明了其在生成城市街景分割数据集方面的有效性，在域内（少样本和全监督）设置以及域泛化任务中均优于基线和最优方法，尤其是在恶劣天气和变化光照等挑战性条件下，进一步凸显了其优越性。
英文原文摘要 1264 字符
This paper addresses the challenge of data scarcity in semantic segmentation by generating datasets through text-to-image (T2I) generation models, reducing image acquisition and labeling costs. Segmentation dataset generation faces two key challenges: 1) aligning generated samples with the target domain and 2) producing informative samples beyond the training data. Fine-tuning T2I models can help generate samples aligned with the target domain. However, it often overfits and memorizes training data, limiting their ability to generate diverse and well-aligned samples. To overcome these issues, we propose Concept-Aware LoRA (CA-LoRA), a novel fine-tuning approach that selectively identifies and updates only the weights associated with necessary concepts (e.g., style or viewpoint) for domain alignment while preserving the pretrained knowledge of the T2I model to produce informative samples. We demonstrate its effectiveness in generating datasets for urban-scene segmentation, outperforming baseline and state-of-the-art methods in in-domain (few-shot and fully-supervised) settings, as well as in domain generalization tasks, especially under challenging conditions such as adverse weather and varying illumination, further highlighting its superiority.