CCAD: Compressed Global Feature Conditioned Anomaly Detection

TL;DR

CCAD addresses industrial anomaly detection under scarce anomalous data by conditioning a diffusion-based reconstruction on compressed global features. It introduces a two-stage global feature compression (coarse and fine banks) and three CCAD variants (CCAD(F/C/V)) to balance accuracy and efficiency, achieving superior AUC and faster convergence across multiple datasets. The approach is complemented by re-annotations of DAGM-2007 and extensive fair-comparison experiments, demonstrating robustness to domain shift and improved pixel-level localization. Together, these contributions offer a practical, scalable solution for high-stakes industrial AD with accessible code and data.

Abstract

Anomaly detection holds considerable industrial significance, especially in scenarios with limited anomalous data. Currently, reconstruction-based and unsupervised representation-based approaches are the primary focus. However, unsupervised representation-based methods struggle to extract robust features under domain shift, whereas reconstruction-based methods often suffer from low training efficiency and performance degradation due to insufficient constraints. To address these challenges, we propose a novel method named Compressed Global Feature Conditioned Anomaly Detection (CCAD). CCAD synergizes the strengths of both paradigms by adapting global features as a new modality condition for the reconstruction model. Furthermore, we design an adaptive compression mechanism to enhance both generalization and training efficiency. Extensive experiments demonstrate that CCAD consistently outperforms state-of-the-art methods in terms of AUC while achieving faster convergence. In addition, we contribute a reorganized and re-annotated version of the DAGM 2007 dataset with new annotations to further validate our method's effectiveness. The code for reproducing main results is available at https://github.com/chloeqxq/CCAD.

Figures (12)

• Figure 1: An overview of Diffusion Modules (DM) and Conditioned Diffusion Modules (CDM): (a) The vanilla DM operates without any condition. (b) A single sample $\mathbf{x}_0$ is used as the condition (c) Compressed vectors $\mathcal{B}$ representing the distribution of a whole dataset are served as the condition.
• Figure 2: The framework of CCAD(F). Our method consists of two main part: Global Feature Compression Block (GFCB) and Global feature Conditioned Diffusion Module (GCDM). FFB denotes Fine Feature Bank, and GCB denotes Global Feature Conditioned Block.
• Figure 3: Architecture of Global feature Conditioned Blocks.
• Figure 4: Qualitative example visualization.
• Figure 5: AUC on MVTec with different $\xi$.