Bi-Grid Reconstruction for Image Anomaly Detection

2025-04-01

AI Summary

In fine-grained image anomaly detection—where only normal samples are available for training—existing methods suffer from weak boundary discrimination and severe identity shortcut interference. To address these issues, this paper proposes a Dual-Continual Grid Collaborative Modeling framework. Our key contributions are: (1) the first Dual-Grid Feature Warehouse, explicitly modeling normal and abnormal feature distributions in parallel; (2) an abnormal-feature grid that dynamically refines the normal decision boundary, mitigating identity shortcuts; and (3) a Feature Block Pasting (FBP) module that synthesizes diverse anomalous patterns directly in feature space. Built upon a self-supervised reconstruction paradigm, our approach integrates dual-grid decoupled representations with a multi-class shared single-model architecture. Extensive experiments on MVTec-AD, VisA, and GoodsAD benchmarks demonstrate significant improvements in fine-grained anomaly localization and subtle-difference discrimination, consistently outperforming state-of-the-art unsupervised and self-supervised methods.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Unsupervised & Self-Supervised Learning
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

In image anomaly detection, significant advancements have been made using un- and self-supervised methods with datasets containing only normal samples. However, these approaches often struggle with fine-grained anomalies. This paper introduces extbf{GRAD}: Bi- extbf{G}rid extbf{R}econstruction for Image extbf{A}nomaly extbf{D}etection, which employs two continuous grids to enhance anomaly detection from both normal and abnormal perspectives. In this work: 1) Grids as feature repositories that improve generalization and mitigate the Identical Shortcut (IS) issue; 2) An abnormal feature grid that refines normal feature boundaries, boosting detection of fine-grained defects; 3) The Feature Block Paste (FBP) module, which synthesizes various anomalies at the feature level for quick abnormal grid deployment. GRAD's robust representation capabilities also allow it to handle multiple classes with a single model. Evaluations on datasets like MVTecAD, VisA, and GoodsAD show significant performance improvements in fine-grained anomaly detection. GRAD excels in overall accuracy and in discerning subtle differences, demonstrating its superiority over existing methods.

Problem

• Improves fine-grained anomaly detection in images
• Mitigates Identical Shortcut issue with dual-grid reconstruction
• Enhances defect detection via feature boundary refinement

Innovation

• Bi-Grid Reconstruction enhances anomaly detection
• Feature Block Paste synthesizes feature-level anomalies
• Dual-grid system refines normal feature boundaries