CostFilter-AD: Enhancing Anomaly Detection through Matching Cost Filtering

2025/05/02 by Zhe Zhang, Meng Cai, Zhang, Zhe +9 · 1 citation
Computer Science · #Anomaly Detection Techniques and Applications #Artificial Intelligence (cs.AI) #Computer Vision and Pattern Recognition (cs.CV) #Data Stream Mining Techniques #FOS: Computer and information sciences #FOS: Electrical engineering #Image and Video Processing (eess.IV) #Network Security and Intrusion Detection #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.2505.01476

openalex publication_date 2025/05/02 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Unsupervised anomaly detection (UAD) seeks to localize the anomaly mask of an input image with respect to normal samples. Either by reconstructing normal counterparts (reconstruction-based) or by learning an image feature embedding space (embedding-based), existing approaches fundamentally rely on image-level or feature-level matching to derive anomaly scores. Often, such a matching process is inaccurate yet overlooked, leading to sub-optimal detection. To address this issue, we introduce the concept of cost filtering, borrowed from classical matching tasks, such as depth and flow estimation, into the UAD problem. We call this approach \em CostFilter-AD. Specifically, we first construct a matching cost volume between the input and normal samples, comprising two spatial dimensions and one matching dimension that encodes potential matches. To refine this, we propose a cost volume filtering network, guided by the input observation as an attention query across multiple feature layers, which effectively suppresses matching noise while preserving edge structures and capturing subtle anomalies. Designed as a generic post-processing plug-in, CostFilter-AD can be integrated with either reconstruction-based or embedding-based methods. Extensive experiments on MVTec-AD and VisA benchmarks validate the generic benefits of CostFilter-AD for both single- and multi-class UAD tasks. Code and models will be released at https://github.com/ZHE-SAPI/CostFilter-AD.

Citations

• A Unified Anomaly Synthesis Strategy with Gradient Ascent for Industrial Anomaly Detection and Localization
• Cross-Modal Learning for Anomaly Detection in Complex Industrial Process: Methodology and Benchmark
• Dinomaly: The Less Is More Philosophy in Multi-Class Unsupervised Anomaly Detection
• AnomalyDINO: Boosting Patch-based Few-shot Anomaly Detection with DINOv2
• MambaAD: Exploring State Space Models for Multi-class Unsupervised Anomaly Detection
• Text-Guided Variational Image Generation for Industrial Anomaly Detection and Segmentation
• RealNet: A Feature Selection Network with Realistic Synthetic Anomaly for Anomaly Detection
• A Survey on Visual Anomaly Detection: Challenge, Approach, and Prospect
• Unified Domain Adaptive Semantic Segmentation
• Hierarchical Vector Quantized Transformer for Multi-class Unsupervised Anomaly Detection
• DIFT: Dynamic Iterative Field Transforms for Memory Efficient Optical Flow
• SimpleNet: A Simple Network for Image Anomaly Detection and Localization
• Diffusion Probabilistic Model Made Slim
• PNI : Industrial Anomaly Detection using Position and Neighborhood Information
• DeSTSeg: Segmentation Guided Denoising Student-Teacher for Anomaly Detection
• SPot-the-Difference Self-Supervised Pre-training for Anomaly Detection and Segmentation
• Geo-NI: Geometry-aware Neural Interpolation for Light Field Rendering
• A Unified Model for Multi-class Anomaly Detection
• Rethinking Depth Estimation for Multi-View Stereo: A Unified Representation
• DRAEM -- A discriminatively trained reconstruction embedding for surface\n anomaly detection
• CFLOW-AD: Real-Time Unsupervised Anomaly Detection with Localization via Conditional Normalizing Flows
• Towards Total Recall in Industrial Anomaly Detection
• Emerging Properties in Self-Supervised Vision Transformers
• Self-supervised Learning of Depth Inference for Multi-view Stereo
• EfficientNetV2: Smaller Models and Faster Training
• Deep One-Class Classification via Interpolated Gaussian Descriptor
• PaDiM: a Patch Distribution Modeling Framework for Anomaly Detection and\n Localization
• Memorizing Normality to Detect Anomaly: Memory-augmented Deep Autoencoder for Unsupervised Anomaly Detection
• CBAM: Convolutional Block Attention Module
• End-to-End Learning of Geometry and Context for Deep Stereo Regression
• 3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation
• ImageNet classification with deep convolutional neural networks
• Image quality assessment: from error visibility to structural similarity
• On Estimation of a Probability Density Function and Mode

Cited by

• One Dinomaly2 Detect Them All: A Unified Framework for Full-Spectrum Unsupervised Anomaly Detection

Related

• Unified Unsupervised Anomaly Detection via Matching Cost Filtering
• UniFormaly: Towards Task-Agnostic Unified Framework for Visual Anomaly Detection
• Divide-and-Assemble: Learning Block-wise Memory for Unsupervised Anomaly Detection
• Not All Regions Are Equal: Attention-Guided Perturbation Network for Industrial Anomaly Detection
• Leveraging Learning Bias for Noisy Anomaly Detection