发表机构
Korea Advanced Institute of Science and Technology (KAIST); Sony Computer Science Laboratories (Sony CSL); The University of Tokyo; National Taiwan Normal University (NTNU)（韩国科学技术院（KAIST）; 索尼计算机科学实验室; 东京大学; 国立台湾师范大学（NTNU））
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
NFAD框架通过建模成像条件干扰变异，在AeBAD-S上实现91.0%图像级AUROC，既提升分布偏移下的异常检测性能，又保持标准基准的竞争力。
AI 中文摘要
工业检测领域的异常检测（AD）近期进展已使标准基准上的性能趋于饱和，但基准性能优异并不一定能迁移到实际部署中，因为这些基准主要在受控采集条件下收集。光照、背景、视角及其他环境因素的变化会使正常样本偏离学习到的正态分布，引发虚假异常响应。针对此类分布偏移下的AD问题，我们通过在特征空间中显式建模成像条件变化带来的干扰变异来解决。无需异常标签或目标域数据，我们的干扰过滤异常检测（NFAD）框架可通过内容保持扰动诱导的匹配特征位移估计干扰子空间，并在推理时抑制其对异常残差的贡献。该子空间支持两个互补分支：全投影用于图像级检测，选择性抑制用于像素级定位，同时保留局部缺陷的证据。在专为采集偏移下AD设计的基准AeBAD-S上，NFAD达到91.0%的图像级AUROC，建立了新的最优性能。值得注意的是，这种鲁棒性并未以牺牲传统AD性能为代价：NFAD在未明确评估分布偏移的标准基准（包括VisA、Real-IAD和MVTec AD）上仍具竞争力。这些结果表明，显式抑制此类干扰变异可提升分布偏移下的AD性能，同时保留标准设置下的强性能。
英文摘要
Recent advances in anomaly detection (AD) for industrial inspection have pushed performance on standard benchmarks toward saturation. However, strong benchmark performance does not necessarily translate to real-world deployment, as these benchmarks are primarily collected under controlled acquisition conditions. Changes in illumination, background, viewpoint, and other environmental factors can shift normal samples away from the learned normal distribution and cause false anomaly responses. We address AD under such distribution shifts by explicitly modeling nuisance variation from changing imaging conditions in feature space. Without anomaly labels or target-domain data, our Nuisance-Filtered Anomaly Detection (NFAD) framework estimates a nuisance subspace from matched feature displacements induced by content-preserving perturbations and suppresses its contribution to anomaly residuals at inference. The same subspace supports two complementary branches: full projection for image-level detection and selective suppression for pixel-level localization, preserving evidence of localized defects. On AeBAD-S, a benchmark specifically designed for AD under acquisition shifts, NFAD achieves 91.0\% image-level AUROC, establishing a new state of the art. Notably, this robustness does not come at the expense of conventional AD performance: NFAD remains competitive on standard benchmarks that do not explicitly evaluate distribution shift, including VisA, Real-IAD, and MVTec AD. These results show that explicitly suppressing such nuisance variation improves AD under distribution shift while preserving strong performance in standard settings.
