Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces SwinAD, a multi-stage reconstruction framework designed for unsupervised anomaly detection in industrial settings, addressing the limitations of existing methods that struggle with multi-class scenarios. By leveraging a frozen pretrained Swin Transformer V2 encoder alongside a feature diversity-preserving reconstruction decoder, the method enhances the discriminative capability between normal and anomalous regions. Experimental results on benchmarks like MVTec AD show that SwinAD significantly improves both image-level performance and pixel-level localization accuracy, particularly excelling in pixel-level average precision (AP) metrics.
SwinAD achieves unprecedented pixel-level localization accuracy in multi-class unsupervised anomaly detection by combining hierarchical features with diverse reconstruction strategies.
Industrial anomaly detection aims to identify and localize defective regions without relying on exhaustive annotations of all possible defect types. Although recent unsupervised methods have achieved strong performance, most are primarily designed for single-class settings and often struggle in multi-class scenarios, where diverse normal patterns may lead to over-generalization and reduce the discriminative capability between normal and anomalous regions. In this paper, we propose SwinAD, a reconstruction-based framework for multi-class unsupervised anomaly detection that leverages a frozen pretrained Swin Transformer V2 encoder and a feature diversity-preserving reconstruction decoder. The hierarchical encoder provides semantically rich multi-scale features, while stage-wise bottleneck modules with dropout prevent trivial identity mapping and encourage robust reconstruction of normal patterns. To further improve localization, we introduce a feature diversity-preserving reconstruction framework that maintains complementary reconstruction hypotheses instead of relying on a single decoding branch. The discrepancies between encoder features and the two reconstructed features are then aggregated across multiple scales to produce the final anomaly map. Experiments conducted on three industrial anomaly detection benchmarks, including MVTec AD, VisA, and Real-IAD, demonstrate that SwinAD achieves competitive image-level performance and strong pixel-level localization accuracy, with particularly notable improvements in pixel-level AP and 1 on MVTec AD. These results indicate that combining hierarchical Swin features with diverse multi-scale reconstruction substantially improve pixel-level localization in multi-class unsupervised anomaly setting.