Beyond Dents and Scratches: Logical Constraints in Unsupervised Anomaly Detection and Localization

Abstract

The unsupervised detection and localization of anomalies in natural images is an intriguing and challenging problem. Anomalies manifest themselves in very different ways and an ideal benchmark dataset for this task should contain representative examples for all of them. We find that existing datasets are biased towards local structural anomalies such as scratches, dents, or contaminations. In particular, they lack anomalies in the form of violations of logical constraints, e.g., permissible objects occurring in invalid locations. We contribute a new dataset based on industrial inspection scenarios that evenly covers both types of anomalies. We provide pixel-precise ground truth data for each anomalous region and define a generalized evaluation metric that addresses localization ambiguities that can arise for logical anomalies. Furthermore, we propose a novel algorithm that improves over the state of the art in the joint detection of structural and logical anomalies. It consists of a local and a global network branch. The first one inspects confined regions independent of their spatial locations in the input image and is primarily responsible for the detection of entirely new local structures. The second one learns a globally consistent representation of the training data through a bottleneck that enables the detection of violations of long-range dependencies, a key characteristic of many logical anomalies. We perform extensive evaluations on our new dataset to corroborate our claims.

Deep Industrial Image Anomaly Detection: A Survey

“…For example, an automaker manufactures several types of workpieces but does not produce fruit. Current popular IAD datasets, like MVTec AD [5] and MVTec LOCO [6], consist of numerous classes but not multiple domains.…”

“…Since the release of MVTec 3D-AD [6] dataset, several papers have focused on anomaly detection in 3D industrial images. Bergmann [134] introduces a teacher-student model for 3D anomaly detection.…”

GRD-Net: Generative-Reconstructive-Discriminative Anomaly Detection with Region of Interest Attention Module

“…Knowledge distillation techniques are also widely used in anomaly detection tasks, especially when we are dealing with large images, as in the work of Paul et al [32]. Tis matter is examined in the work also written by Paul Bergmann et al [33], in which anomalies are divided into logical and structural. Noteworthy is also the knowledge distillationbased work of Kilian Batzner et al [34] where processing time plays a central role in the problem defnition because more and more often lots of real-time applications use unsupervised machine learning algorithms for anomaly detection tasks.…”

Zero-Shot and Few-Shot Learning With Knowledge Graphs: A Comprehensive Survey

“…In the field of computer vision, unsupervised anomaly detection (AD) [5,13,15] aims to identify abnormal images and locate anomalous regions using a model trained solely on anomaly-free images. It is widely used in industrial defect detection [1,2,11]. Most previous methods have centered on training dedicated models for each category, relying on a vast collection of normal images as references [14,16].…”