FastRef：面向少样本工业异常检测的快速原型精炼
FastRef: Fast Prototype Refinement for Few-shot Industrial Anomaly Detection
把推理阶段的查询特征通过可优化传输矩阵做线性重构，并用最优传输约束抑制异常重构，这套「测试时原型精炼」思路可直接迁移到任何基于原型的少样本检测/分割流程中，无需重训练即可提升表征代表性。
原文依据 we employ optimal transport to measure while minimize the gap between prototypes and their enhanced counterparts for anomaly suppression
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论少样本工业异常检测（FS-IAD）对在数据稀缺环境下运行的实际自动检测系统构成关键挑战。现有方法主要聚焦于从有限的正常图像中获取原型，却忽视了系统性地引入查询图像的统计信息以增强原型的代表性。为解决这一问题，我们提出 FastRef，一种面向 FS-IAD 的新颖且高效的原型精炼框架。该方法在推理阶段通过迭代的两阶段过程运行：（1）从查询特征向增强原型进行特征迁移；（2）通过将增强原型与其正常对应原型对齐来抑制异常。特征迁移通过以可优化传输矩阵从原型对查询特征进行线性重构来实现，而异常抑制则针对 FS-IAD 中的一个关键观察：与拥有充足正常原型的传统 IAD 不同，有限样本设定使特征迁移中出现异常重构的可能性更高。因此，我们采用最优传输来度量并最小化原型与其增强对应原型之间的差距，以实现异常抑制。为进行全面的评估，我们将 FastRef 与三种具有竞争力的基于原型的 FS-IAD 方法（PatchCore、WinCLIP 和 AnomalyDINO）集成。在 MVTec、ViSA、MPDD 和 RealIAD 四个基准数据集上的大量实验表明，我们的方法在 1/2/4 样本设定下兼具有效性与高效性。
英文原文摘要 1556 字符
Few-shot industrial anomaly detection (FS-IAD) presents a critical challenge for practical automated inspection systems operating in data-scarce environments. While existing approaches predominantly focus on obtaining prototypes from limited normal images, they neglect to systematically incorporate statistics of query image to enhance prototype representativeness. To address this issue, we propose FastRef, a novel and efficient prototype refinement framework for FS-IAD. Our method operates through an iterative two-stage process during inference: (1) characteristic transfer from query features to the enhanced prototypes, and (2) anomaly suppression by aligning the enhanced prototypes with their normal counterparts. The characteristic transfer is achieved through linear reconstruction of query features from prototypes with an optimizable transport matrix, while the anomaly suppression addresses a key observation in FS-IAD that unlike conventional IAD with abundant normal prototypes, the limited-sample setting makes anomaly reconstruction more probable in characteristic transfer. Therefore, we employ optimal transport to measure while minimize the gap between prototypes and their enhanced counterparts for anomaly suppression. For comprehensive evaluation, we integrate FastRef with three competitive prototype-based FS-IAD methods: PatchCore, WinCLIP, and AnomalyDINO. Extensive experiments across four benchmark datasets of MVTec, ViSA, MPDD and RealIAD demonstrate both the effectiveness and efficiency of our approach under 1/2/4-shots.