Abstract:Anomaly detection is a crucial means to detect unbalanced and multiclass abnormal data in industrial products. The existing anomaly detection model is usually suitable for products with a simple or regular background texture. However, there is a lack of a universal surface defect anomaly detection framework for railway products when facing the challenges: the complex background texture, huge intraclass variation, uneven distribution of interclass, and weak texture-featured defects. To fill this gap, an innovative generative adversarial network based on adaptive pyramid graph (APG) and variation residuals (APGVR-GAN) is proposed, aiming to improve the robustness of anomaly detection in railway products and other complex industrial supplies. First, the APG module is embedded in the encoder–decoder–encoder pattern, capturing the correlation description between neighbor regions, which is utilized to enhance the detection of abnormal defects with weak texture. Next, the variation residual module is employed to enhance the expression of various normal samples in the latent space and improve the identification ability for abnormal samples. Then, the dual-probability prototype loss is proposed to make different normal samples have more concentrated expression and more similar probability distribution centers in latent space. Finally, an adaptive focal-gate loss and a regularized log-likelihood loss are designed to overcome the imbalance problem in training samples with different background information. The effectiveness of the model is verified on three new railway datasets and three other industrial public datasets. The results show that in anomaly detection, APGVR-GAN on the three railway datasets, Northeastern University (NEU)-the rail surface defects dataset (RSDD)-1 dataset, NEU-RSDD-1 dataset, and RSDD area under curve (AUC) [that is the receiver operating characteristic curve (ROC)], can reach 0.89, 0.81, and 0.87, and its detection accuracy (ACC) reached 0.97, 0.96, and 0.92, which is much higher than other comparative anomaly detection algorithms. In the segmentation experiments on the RSDD dataset, compared with other known rail defect detection methods, the overall evaluation metric F-Score is 83.70 which is also better than other methods.

MRSA-Net: Anomaly Detection of Metro Stations Based on Residual Attention Network with Path Aggregation and Efficient Representation

Segmenting Remote Sensing Anomalies at Instance-level Via Anomaly Map Guided Adaptation

+2cmIntercity Rail Transit Platform Anomaly Detection Using Door Tracking Based Key Frame Extraction and AnoDet Network

Object-Guided and Motion-Refined Attention Network for Video Anomaly Detection

Unified Model Based on Reinforced Feature Reconstruction for Metro Track Anomaly Detection

MSCA: an Unsupervised Anomaly Detection System for Network Security in Backbone Network.

A Diffusion-Based Framework for Multi-Class Anomaly Detection

Attention Based CNN-LSTM Network for Anomaly Pattern Classification of Multivariate Time Series

Anomaly-GAN: A data augmentation method for train surface anomaly detection

An Adaptive Pyramid Graph and Variation Residual-Based Anomaly Detection Network for Rail Surface Defects

AnomalyNCD: Towards Novel Anomaly Class Discovery in Industrial Scenarios

Multi-scale Spatial-temporal Interaction Network for Video Anomaly Detection

A Novel MAE-Based Self-Supervised Anomaly Detection and Localization Method

AnomalyNet: a spatiotemporal motion-aware CNN approach for detecting anomalies in real-world autonomous surveillance

A Novel Method for Road Anomaly Objects Detection in the Traffic Environment With Multi-Mechanism Fusion

Attention-Guided Perturbation for Unsupervised Image Anomaly Detection

Improved LSTM-based Time-Series Anomaly Detection in Rail Transit Operation Environments

Online Growing Neural Gas for Anomaly Detection in Changing Surveillance Scenes

Structural Teacher-Student Normality Learning for Multi-Class Anomaly Detection and Localization

Influence-aware Attention Networks for Anomaly Detection in Surveillance Videos

MSTAD: A masked subspace-like transformer for multi-class anomaly detection