Visualization for Multivariate Gaussian Anomaly Detection in Images
This work addresses anomaly detection for image analysis, but it is incremental as it builds on existing PaDiM methods.
The paper tackled anomaly detection in images by proposing a simplified PaDiM variant using a multivariate Gaussian distribution and whitening transformation for heatmap generation, evaluated on the MVTec-AD dataset to highlight visual model validation insights.
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.