Style-Aware Cross Domain Few-Shot Anomaly Detection

Few-shot anomaly detection involves using a limited number of normal samples to detect and localize anomalies. While existing methods for few-shot anomaly detection are primarily applied within the same domain, research on cross-domain scenarios remains limited. To address the problem of cross-domain few-shot anomaly detection, a cross-domain few-shot anomaly detection method based on style memory is proposed. This method includes two modules. The first module defines domain differences in terms of differences in style features and employs a style memory module to store intra-domain statistical features from the source domain. These features are then integrated into the target domain to mitigate the impact of domain differences and enhance few-shot detection performance. Then, a contrastive learning strategy is introduced to distinguish enhanced query features of different categories, further enhancing the model’s ability to represent different categories in the source domain. Experiments on cross-domain few-shot anomaly detection tasks between the MVTec AD, MPDD and MVTec LOCO-AD datasets show that the proposed method achieves the best performance compared to the represntative methods, with an image AUROC indicator increase of 3% to 5%. Moreover, the proposed model still has the best performance on the few-shot anomaly detection task under the intra-domain setting.