Reconstruction-based methods are a cornerstone of unsupervised image anomaly detection, but they remain vulnerable to \emph{outlier leakage}, where standard mean squared error (MSE) loss drives the model to faithfully reconstruct anomalous patterns. We propose a Non-linear Reconstruction Loss that applies a sigmoid-based squashing function to suppress high-magnitude features, preventing outliers from dominating optimization while preserving sensitivity to normal patterns.
The paper introduces a training‑free anomaly detector that simultaneously handles structural and logical defects by calibrating heterogeneous anomaly cues with statistics from normal images. This calibration aligns frozen representations, allowing their fusion without extra training or part‑level supervision. The resulting method achieves state‑of‑the‑art AUROC scores on MVTec‑LOCO and remains competitive on MVTec‑AD.
By Changyi Li, Miao Yu, Kai Dong, Yu Xiao
The paper introduces a variational template matching framework for anomaly detection in patterned structures, representing anomaly templates as transformed instances and using normalized cross‑correlation across the transformation space. It enhances robustness by adding a density‑based statistical anomaly score derived from local intensity distributions via kernel density estimation, which captures distributional concentration and tail behavior more effectively than histogram methods. The structural and statistical cues are fused in a unified formulation, and experiments on biological cell images show the method outperforms classical baselines and rivals ResNet‑50 while remaining fully training‑free and providing explicit localization.
By Qinwu Xu, Yifan Jiang
VerdictAD is a training‑free framework that enhances existing few‑shot anomaly detectors by adding a typed decision model to evaluate query images and their local regions against a small set of normal references. It introduces two modules—Global Verdict Refinement (GVR) and Local Verdict Refinement (LVR)—which refine overall anomaly scores and anomaly maps, respectively, without generating text. When combined with a simple DINOv2‑based detector, VerdictAD consistently improves performance on MVTec AD, VisA, and MVTec LOCO AD across one‑shot and few‑shot settings.
arXiv:2510. 06505v2 Announce Type: replace-cross Abstract: Out-of-distribution (OOD) detection plays a crucial role in ensuring the robustness of machine learning systems deployed in real-world applications.
By Momin Abbas, Ali Falahati, Hossein Goli, Mohammad Mohammadi Amiri
arXiv:2510. 24043v4 Announce Type: replace Abstract: This paper presents Two-Stage LKPLO, a novel multi-stage outlier detection framework that overcomes the coexisting limitations of conventional projection-based methods: their reliance on a fixed statistical metric and their assumption of a single data structure.
By Akira Tamamori
arXiv:2606. 28970v1 Announce Type: cross Abstract: Unsupervised tabular anomaly detection requires methods that are accurate, robust across heterogeneous datasets, and computationally efficient.
By Quanling Zhao, Jiaying Yang, Ye Tian, Josh Victoria, Zhijun Wang, Pietro Mercati, Onat Gungor, Tajana Rosing
arXiv:2609.37314v1 Announce Type: new
Abstract: Synthetic anomaly generation helps expand industrial anomaly datasets when real defects are scarce or unavailable. Existing approaches lie at two extre...
By Abhay Kumar Das, Rajesh Gangireddy, Ashwin Vaidya, Samet Akcay
The paper identifies a problem in multi‑view anomaly detection called cross‑view information leakage, where fusing multiple inspection views can cause normal features to mask anomalies during reconstruction. To address this, the authors propose GLAD, a framework that uses a Global‑Local Attention Driven approach, combining vision foundation model features with two fusion modules: Multi‑view Merging Attention for local, weighted fusion and Object‑Guided Attention for global context aggregation. Experiments on Real‑IAD and MANTA‑Tiny demonstrate that GLAD outperforms existing methods across various metrics, underscoring the importance of restricting information flow to preserve the reconstruction gap.
By Shang-Fu Chen, Kuan-Chuan Peng, Jhih-Ciang Wu, Wen-Huang Cheng, Kai-Lung Hua
The paper introduces Contrastive Dual Gaussian Processes (CDGP), a weakly supervised approach for anomaly segmentation in industrial visual inspection. CDGP models normal and anomaly predictive distributions over dense tokens, using a posterior-dominance statistic that normalizes predictive-mean differences by joint uncertainty to provide spatial evidence and image-level confidence. The method achieves top performance on MVTec AD~2, KSDD2, and VisA datasets without requiring pixel-level annotations or test-time fitting.
The paper challenges the common assumption that RGB and point cloud data contribute equally to zero‑shot multimodal anomaly detection. It shows that point clouds are more reliable under category shift and introduces WOOPS, a framework that enhances point cloud features with a Multi‑view Information Decoupling module and calibrates modality contributions via a Modality Reliability Calibration module. Experiments demonstrate that WOOPS achieves top performance on new stringent metrics in both unimodal and multimodal settings, and that point cloud information also benefits RGB‑only inference.
By Chenglin Ye, Lupeng Liu, Dongbo Yu, Jun Xiao, Yunbiao Wang
arXiv:2608.29112v1 Announce Type: new
Abstract: Recent advances in anomaly detection (AD) for industrial inspection have pushed performance on standard benchmarks toward saturation. However, strong b...
By Dat Cao, Son Nghiem, Phan Nguyen, Jun Rekimoto, Jhih-Ciang Wu