Domain-independent detection of known anomalies

TL;DR

This paper introduces a new hybrid task combining domain generalization and anomaly detection on sparse classes, along with modified datasets and novel embedding-based methods, achieving improved detection performance.

Contribution

It proposes the hybrid task of domain generalization on sparse classes, introduces new datasets, and develops two embedding-based methods, SEMLP and Labeled PatchCore.

Findings

SEMLP achieves an average image-level AUROC of 87.2%.

The new datasets facilitate further research in industrial anomaly detection.

The approach outperforms existing methods like MIRO.

Abstract

One persistent obstacle in industrial quality inspection is the detection of anomalies. In real-world use cases, two problems must be addressed: anomalous data is sparse and the same types of anomalies need to be detected on previously unseen objects. Current anomaly detection approaches can be trained with sparse nominal data, whereas domain generalization approaches enable detecting objects in previously unseen domains. Utilizing those two observations, we introduce the hybrid task of domain generalization on sparse classes. To introduce an accompanying dataset for this task, we present a modification of the well-established MVTec AD dataset by generating three new datasets. In addition to applying existing methods for benchmark, we design two embedding-based approaches, Spatial Embedding MLP (SEMLP) and Labeled PatchCore. Overall, SEMLP achieves the best performance with an average image-level AUROC of 87.2 % vs. 80.4 % by MIRO. The new and openly available datasets allow for further research to improve industrial anomaly detection.