Isolating to Harness: Cross-Division Distillation for Fully Unsupervised Anomaly Detection

2025/08/25 by Xinyue Liu, Liu, Xinyue, Jianyuan Wang +5
Computer Science · #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #cs.CV

paper · pdf · doi:10.48550/arxiv.2508.18007

Accepted by TCSVT

arxiv created 2026/07/30 · arxiv updated 2026/07/31

Abstract

Fully Unsupervised Anomaly Detection (FUAD) addresses the practical scenario where training data is contaminated with unlabeled anomalies. This setting critically challenges conventional Unsupervised Anomaly Detection (UAD) methods, as they tend to misinterpret training anomalies as normal patterns, leading to false negatives. Although filtering anomalies from the training set is a common countermeasure, it inevitably discards valuable data and degrades the model's representation of normality. To overcome this dilemma, we propose an "isolating to harness" strategy, which isolates the influence of anomalies within specialized divisions and then leverages cross-division collaboration to generate robust pseudo supervision. We materialize this idea via a novel Cross-Division Distillation framework based on the widely studied Reverse Distillation paradigm. CDD first partitions the data into divisions with reduced anomaly ratios to train division-specific students. It then aggregates pseudo-normal features generated by each division-specific student for samples from other data divisions to guide a global student towards a robust anomaly-free representation. Experimental results on noisy versions of multiple AD datasets demonstrate that our method achieves significant performance improvements over the baseline. Code is available at https://github.com/hito2448/CDD.

Citations

• TailedCore: Few-Shot Sampling for Unsupervised Long-Tail Noisy Anomaly Detection
• Unlocking the Potential of Reverse Distillation for Anomaly Detection
• FUN-AD: Fully Unsupervised Learning for Anomaly Detection with Noisy Training Data
• Dinomaly: The Less Is More Philosophy in Multi-Class Unsupervised Anomaly Detection
• Real-IAD: A Real-World Multi-View Dataset for Benchmarking Versatile Industrial Anomaly Detection
• RealNet: A Feature Selection Network with Realistic Synthetic Anomaly for Anomaly Detection
• Vision Transformers Need Registers
• ReContrast: Domain-Specific Anomaly Detection via Contrastive Reconstruction
• ReConPatch : Contrastive Patch Representation Learning for Industrial Anomaly Detection
• PNI : Industrial Anomaly Detection using Position and Neighborhood Information
• Visual Anomaly Detection Via Partition Memory Bank Module and Error Estimation
• SPot-the-Difference Self-Supervised Pre-training for Anomaly Detection and Segmentation
• A Unified Model for Multi-class Anomaly Detection
• DRAEM -- A discriminatively trained reconstruction embedding for surface\n anomaly detection
• Towards Total Recall in Industrial Anomaly Detection
• CutPaste: Self-Supervised Learning for Anomaly Detection and Localization
• Multiresolution Knowledge Distillation for Anomaly Detection
• PaDiM: a Patch Distribution Modeling Framework for Anomaly Detection and\n Localization
• An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
• FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence
• MixMatch: A Holistic Approach to Semi-Supervised Learning
• GANomaly: Semi-Supervised Anomaly Detection via Adversarial Training
• Wide Residual Networks

Related

• Unlocking the Potential of Reverse Distillation for Anomaly Detection
• Generalist Multi-Class Anomaly Detection via Distillation to Two Heterogeneous Student Networks
• Advancing Pre-trained Teacher: Towards Robust Feature Discrepancy for Anomaly Detection
• Tuned Reverse Distillation: Enhancing Multimodal Industrial Anomaly Detection with Crossmodal Tuners
• Leveraging Learning Bias for Noisy Anomaly Detection