Abstract:Anomaly Detection (AD) in images is a fundamental computer vision problem and refers to identifying images and/or image substructures that deviate significantly from the norm. Popular AD algorithms commonly try to learn a model of normality from scratch using task specific datasets, but are limited to semi-supervised approaches employing mostly normal data due to the inaccessibility of anomalies on a large scale combined with the ambiguous nature of anomaly appearance. We follow an alternative approach and demonstrate that deep feature representations learned by discriminative models on large natural image datasets are well suited to describe normality and detect even subtle anomalies in a transfer learning setting. Our model of normality is established by fitting a multivariate Gaussian (MVG) to deep feature representations of classification networks trained on ImageNet using normal data only. By subsequently applying the Mahalanobis distance as the anomaly score we outperform the current state of the art on the public MVTec AD dataset, achieving an Area Under the Receiver Operating Characteristic curve of 95.8 ± 1.2% (mean ± SEM) over all 15 classes. We further investigate why the learned representations are discriminative to the AD task using Principal Component Analysis. We find that the principal components containing little variance in normal data are the ones crucial for discriminating between normal and anomalous instances. This gives a possible explanation to the often subpar performance of AD approaches trained from scratch using normal data only. By selectively fitting a MVG to these most relevant components only, we are able to further reduce model complexity while retaining AD performance. We also investigate setting the working point by selecting acceptable False Positive Rate thresholds based on the MVG assumption. Code is publicly available at https://github.com/ORippler/gaussian-ad-mvtec.

Normalizing Flow-Based Probability Distribution Representation Detector for Hyperspectral Anomaly Detection.

Exploring the Intrinsic Probability Distribution for Hyperspectral Anomaly Detection

Exploiting the Sparse Characteristics in Probabilistic Feature Space for Hyperspectral Anomaly Detection

Spatial Anomaly Detection in Hyperspectral Imaging Using Optical Neural Networks

Deep Low-Rank Prior for Hyperspectral Anomaly Detection

Local-density-based Potential Dictionary Construction for Low Rank Representation in Hyperspectral Anomaly Detection

Iterative Graph Propagation for Hyperspectral Anomaly Detection

Hyperspectral Anomaly Detection with Auto-Encoder and Independent Target

Low-Rank and Sparse Decomposition With Mixture of Gaussian for Hyperspectral Anomaly Detection

FastFlow: Unsupervised Anomaly Detection and Localization via 2D Normalizing Flows

Hierarchical Gaussian Mixture Normalizing Flow Modeling for Unified Anomaly Detection

Enhanced Tensor Low-Rank Representation Learning for Hyperspectral Anomaly Detection

Self-Paced Probabilistic Collaborative Representation for Anomaly Detection of Hyperspectral Images

A Weighted Collaborated Representation Anomaly Detector for Hyperspectral Image

Graph and Sparsity Regularized Decomposition Model for Hyperspectral Anomaly Detection

Anomaly detection based on low probability detection for hyper-spectral image

CFLOW-AD: Real-Time Unsupervised Anomaly Detection with Localization via Conditional Normalizing Flows

Modeling the Distribution of Normal Data in Pre-Trained Deep Features for Anomaly Detection

Understanding Likelihood of Normalizing Flow and Image Complexity through the Lens of Out-of-Distribution Detection

Unsupervised Anomaly Detection and Localization Based on Two-Hierarchy Normalizing Flow