GeneralAD: Anomaly Detection Across Domains by Attending to Distorted Features

TL;DR

GeneralAD is a versatile anomaly detection framework that leverages Vision Transformers and self-supervised pseudo-abnormal sample generation to effectively identify and localize anomalies across diverse domains with minimal adjustments.

Contribution

The paper introduces a novel self-supervised anomaly generation module and an attention-based discriminator tailored for cross-domain anomaly detection using Vision Transformers.

Findings

Achieved state-of-the-art results on six datasets.

Performed on par with best methods on remaining datasets.

Effective in both anomaly localization and detection tasks.

Abstract

In the domain of anomaly detection, methods often excel in either high-level semantic or low-level industrial benchmarks, rarely achieving cross-domain proficiency. Semantic anomalies are novelties that differ in meaning from the training set, like unseen objects in self-driving cars. In contrast, industrial anomalies are subtle defects that preserve semantic meaning, such as cracks in airplane components. In this paper, we present GeneralAD, an anomaly detection framework designed to operate in semantic, near-distribution, and industrial settings with minimal per-task adjustments. In our approach, we capitalize on the inherent design of Vision Transformers, which are trained on image patches, thereby ensuring that the last hidden states retain a patch-based structure. We propose a novel self-supervised anomaly generation module that employs straightforward operations like noise addition and shuffling to patch features to construct pseudo-abnormal samples. These features are fed to an attention-based discriminator, which is trained to score every patch in the image. With this, our method can both accurately identify anomalies at the image level and also generate interpretable anomaly maps. We extensively evaluated our approach on ten datasets, achieving state-of-the-art results in six and on-par performance in the remaining for both localization and detection tasks.