发表机构
Faculty of Computing, Harbin Institute of Technology, Harbin, China University of New South Wales, Sydney, Australia School of Biomedical Engineering, Tsinghua University, Beijing, China School of Artificial Intelligence; Robotics, Hunan University, Changsha, China Big data institute, Central South University, Sydney, Australia DZ-Matrix, Beijing, China Chongqing Research Institute of HIT, Chongqing, China（; ）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对工业异常检测中冷启动场景下正常样本不足、异常样本稀少的问题，提出基于推拉学习范式的即插即用校准框架ArcAD，通过将正常样本投影到超球面并聚类、合成伪异常并利用真实异常优化边界，在多个数据集上显著优于现有方法。
AI 中文摘要
工业异常检测（IAD）在实际制造中经常面临具有挑战性的冷启动瓶颈，其中有限的正常样本无法代表完整的正常分布，且仅有少量异常可用。在这种条件下，现有方法难以形成紧凑的正常边界，也无法有效利用来自罕见缺陷的监督信号。为了解决这一挑战，我们提出了异常校正冷启动异常检测（ArcAD），一个用于基于重建的IAD基线的即插即用校准框架。ArcAD遵循推拉学习范式，在数据稀缺情况下构建紧凑且具有判别性的正常边界。一方面，ArcAD将有限的正常样本投影到超球面上，并将它们拉入多个紧凑的簇中，以最大化对正常流形的覆盖。另一方面，它在超球面上合成伪异常，并利用真实异常向内推动边界，增强异常判别能力。在MVTec-AD、VisA、Real-IAD和MANTA上的大量实验表明，在冷启动条件下，ArcAD在单类和多类设置中均显著优于现有的监督和无监督方法。代码可在以下网址获取：this https URL。
英文摘要
The deployment of Industrial Anomaly Detection (IAD) in real-world manufacturing frequently encounters a challenging cold-start bottleneck, in which limited normal samples fail to represent the full normal distribution and only a few anomalies are available. Under such a regime, existing methods struggle to form compact normal boundaries and fail to effectively exploit supervised signals from rare defects. To address this challenge, we propose Anomaly-Rectified Cold-start AD (ArcAD), a plug-and-play calibration framework for reconstruction-based IAD baselines. ArcAD follows a push-pull learning paradigm to construct a compact and discriminative normal boundary under data scarcity. On the one hand, ArcAD projects limited normal samples onto a hypersphere and pulls them into multiple compact clusters to maximize coverage of the normal manifold. On the other hand, it synthesizes pseudo-anomalies on the hypersphere and leverages real anomalies to push the boundary inward and sharpen anomaly discrimination. Extensive experiments on MVTec-AD, VisA, Real-IAD, and MANTA demonstrate that ArcAD significantly outperforms state-of-the-art supervised and unsupervised methods in both single-class and multi-class settings under cold-start conditions. Code is available at: https://github.com/LGC-AD/ArcAD.
CommentsAccepted to European Conference on Computer Vision (ECCV) 2026
