PaDiM: a Patch Distribution Modeling Framework for Anomaly Detection and Localization

TL;DR

PaDiM is a novel framework that uses pretrained CNN features and Gaussian modeling to effectively detect and localize anomalies in images, outperforming existing methods on industrial datasets.

Contribution

Introduces PaDiM, a new patch distribution modeling approach utilizing CNN features and Gaussian distributions for improved anomaly detection and localization.

Findings

Outperforms state-of-the-art on MVTec AD and STC datasets.

Effective in both detection and localization tasks.

Low complexity suitable for industrial applications.

Abstract

We present a new framework for Patch Distribution Modeling, PaDiM, to concurrently detect and localize anomalies in images in a one-class learning setting. PaDiM makes use of a pretrained convolutional neural network (CNN) for patch embedding, and of multivariate Gaussian distributions to get a probabilistic representation of the normal class. It also exploits correlations between the different semantic levels of CNN to better localize anomalies. PaDiM outperforms current state-of-the-art approaches for both anomaly detection and localization on the MVTec AD and STC datasets. To match real-world visual industrial inspection, we extend the evaluation protocol to assess performance of anomaly localization algorithms on non-aligned dataset. The state-of-the-art performance and low complexity of PaDiM make it a good candidate for many industrial applications.