AnomalyDINO: Boosting Patch-based Few-shot Anomaly Detection with DINOv2

TL;DR

AnomalyDINO is a simple, training-free, vision-only method that leverages DINOv2 features for one-shot and few-shot anomaly detection, outperforming many existing approaches in industrial applications.

Contribution

The paper introduces AnomalyDINO, a novel patch-based, training-free anomaly detection method using DINOv2 features, eliminating the need for fine-tuning or additional data.

Findings

Achieves state-of-the-art one-shot AUROC of 96.6% on MVTec-AD

Outperforms existing methods in few-shot anomaly detection

Simple and fast deployment suitable for industrial use

Abstract

Recent advances in multimodal foundation models have set new standards in few-shot anomaly detection. This paper explores whether high-quality visual features alone are sufficient to rival existing state-of-the-art vision-language models. We affirm this by adapting DINOv2 for one-shot and few-shot anomaly detection, with a focus on industrial applications. We show that this approach does not only rival existing techniques but can even outmatch them in many settings. Our proposed vision-only approach, AnomalyDINO, follows the well-established patch-level deep nearest neighbor paradigm, and enables both image-level anomaly prediction and pixel-level anomaly segmentation. The approach is methodologically simple and training-free and, thus, does not require any additional data for fine-tuning or meta-learning. The approach is methodologically simple and training-free and, thus, does not require any additional data for fine-tuning or meta-learning. Despite its simplicity, AnomalyDINO achieves state-of-the-art results in one- and few-shot anomaly detection (e.g., pushing the one-shot performance on MVTec-AD from an AUROC of 93.1% to 96.6%). The reduced overhead, coupled with its outstanding few-shot performance, makes AnomalyDINO a strong candidate for fast deployment, e.g., in industrial contexts.