"Stones from Other Hills can Polish Jade": Zero-shot Anomaly Image Synthesis via Cross-domain Anomaly Injection

TL;DR

The paper tackles industrial image anomaly detection when domain-specific anomalies are rare by proposing a training-free zero-shot anomaly synthesis framework that leverages abundant cross-domain anomalies. The core idea, Cross-domain Anomaly Injection (CAI), directly injects real cross-domain anomaly patterns into normal target images, supported by a Domain-agnostic Anomaly Dataset (DAAD) of about 6000 real anomalies and a CAI-guided Diffusion Mechanism (CDM) to further extend anomaly diversity via diffusion. Empirical results on MvTec-AD, VisA, and KSDD2 show that CAI yields superior IAD performance and authenticity compared to state-of-the-art ZSAS methods, with DAAD and CDM providing scalable, training-free data foundations and diffusion-based extension. This approach offers a practical, scalable pathway for deploying robust IAD in domains lacking domain-specific anomaly data, while acknowledging current focus on appearance anomalies and suggesting directions for semantic/logical anomaly synthesis.

Abstract

Industrial image anomaly detection (IAD) is a pivotal topic with huge value. Due to anomaly's nature, real anomalies in a specific modern industrial domain (i.e. domain-specific anomalies) are usually too rare to collect, which severely hinders IAD. Thus, zero-shot anomaly synthesis (ZSAS), which synthesizes pseudo anomaly images without any domain-specific anomaly, emerges as a vital technique for IAD. However, existing solutions are either unable to synthesize authentic pseudo anomalies, or require cumbersome training. Thus, we focus on ZSAS and propose a brand-new paradigm that can realize both authentic and training-free ZSAS. It is based on a chronically-ignored fact: Although domain-specific anomalies are rare, real anomalies from other domains (i.e. cross-domain anomalies) are actually abundant and directly applicable to ZSAS. Specifically, our new ZSAS paradigm makes three-fold contributions: First, we propose a novel method named Cross-domain Anomaly Injection (CAI), which directly exploits cross-domain anomalies to enable highly authentic ZSAS in a training-free manner. Second, to supply CAI with sufficient cross-domain anomalies, we build the first Domain-agnostic Anomaly Dataset within our best knowledge, which provides ZSAS with abundant real anomaly patterns. Third, we propose a CAI-guided Diffusion Mechanism, which further breaks the quantity limit of real anomalies and enable unlimited anomaly synthesis. Our head-to-head comparison with existing ZSAS solutions justifies our paradigm's superior performance for IAD and demonstrates it as an effective and pragmatic ZSAS solution.

Figures (7)

• Figure 1: ZSAS flow of Cross-domain Anomaly Injection.
• Figure 2: Comparison of pseudo anomaly images synthesized by SOTA ZSAS solutions and the proposed CAI.
• Figure 3: Detailed procedure of CAI for ZSAS.
• Figure 4: Procedure to build our domain-agnostic anomaly dataset.
• Figure 5: Pseudo anomalies of CAI (row 1) and CDM (row 2-5).