Search

Image anomaly detection is a trending research topic in computer vision. The objective is to build models using available normal samples to detect various abnormal images without depending on real abnormal samples. It has high research significance and value for applications in the detection of defects in product appearance, medical image analysis, hyperspectral image processing, and other fields. This paper proposes an image anomaly detection algorithm based on feature distillation and an autoencoder structure, which uses the feature distillation structure of a dual-teacher network to train the encoder, thus suppressing the reconstruction of abnormal regions. This system also introduces an attention mechanism to highlight the detection objects, achieving effective detection of different defects in product appearance. In addition, this paper proposes a method for anomaly evaluation based on patch similarity that calculates the difference between the reconstructed image and the input image according to different regions of the image, thus improving the sensitivity and accuracy of the anomaly score. This paper conducts experiments on several datasets, and the results show that the proposed algorithm has superior performance in image anomaly detection. It achieves 98.8% average AUC on the SMDC-DET dataset and 98.9% average AUC on the MVTec-AD dataset. [ABSTRACT FROM AUTHOR]

Fatigue failure is a typical failure mode of welded structures. It is of great engineering significance to predict the remaining fatigue life of structures after a certain period of service. In this paper, a two-stage hybrid deep learning approach is proposed only using the response of structures under fatigue loading to predict the remaining fatigue life. In the first stage, a combination of convolutional neural network (CNN), squeeze-and-excitation (SE) block, and long short-term memory (LSTM) network is employed to calculate health indicator values based on the current measured data sequence. In the second stage, a particle filtering-based algorithm is utilized to predict the remaining fatigue life using the previously calculated health indicators. Experimental results on different welded specimens under the same loading conditions demonstrate that the hybrid deep learning approach achieves superior prediction accuracy and generalization ability compared to CNN, LSTM, or CNN-LSTM models in the first stage. Moreover, the average relative deviation between the predicted and actual fatigue life is less than 6% during the final quarter of the crack propagation and fracture stage. [ABSTRACT FROM AUTHOR]