Visualization for Multivariate Gaussian Anomaly Detection in Images

TL;DR

This paper presents a simplified MVG-based method for image anomaly detection that includes a whitening step for visual explanation, validated on the MVTec-AD dataset to improve interpretability.

Contribution

It introduces a whitening transformation in MVG anomaly detection, enabling visual explanations and better model validation in image anomaly detection tasks.

Findings

Whitening improves visual interpretability of the model.

The method effectively detects anomalies on MVTec-AD.

Visualizations aid in diagnosing model issues.

Abstract

This paper introduces a simplified variation of the PaDiM (Pixel-Wise Anomaly Detection through Instance Modeling) method for anomaly detection in images, fitting a single multivariate Gaussian (MVG) distribution to the feature vectors extracted from a backbone convolutional neural network (CNN) and using their Mahalanobis distance as the anomaly score. We introduce an intermediate step in this framework by applying a whitening transformation to the feature vectors, which enables the generation of heatmaps capable of visually explaining the features learned by the MVG. The proposed technique is evaluated on the MVTec-AD dataset, and the results show the importance of visual model validation, providing insights into issues in this framework that were otherwise invisible. The visualizations generated for this paper are publicly available at https://doi.org/10.5281/zenodo.7937978.