Transformer-based multi-scale reconstruction network for defect detection of infrared images

Wei, Changyun, Han, Hui, Wu, Zhichao, Xia, Yu and Ji, Ze ORCID: https://orcid.org/0000-0002-8968-9902 2024. Transformer-based multi-scale reconstruction network for defect detection of infrared images. IEEE Transactions on Instrumentation and Measurement 73 , 5037414. 10.1109/TIM.2024.3481573

PDF - Accepted Post-Print Version Download (6MB)

Abstract

Bottle packaging is extensively used in manufacturing, and inspecting aluminum foil sealing during filling is crucial for ensuring product quality. Traditional Machine vision methods based on supervised learning require extensive annotated data, but the scarcity of defective samples hampers the effectiveness of these methods. To address this challenge, unsupervised learning methods have emerged. Despite their potential, these methods often struggle to accurately learn the distribution of normal samples, resulting in higher rates of false positives and negatives. This paper proposes an unsupervised learning-based approach for anomaly detection in infrared images. Specifically, we construct a Transformer-based multi-scale image reconstruction network (TMIRN) that includes a feature extraction module, a feature fusion module, a reconstruction module, a discriminator network, and an anomaly scoring module. By effectively combining Transformer and CNN techniques, the proposed network excels at capturing both global and local semantic information. Its multi-scale structure accurately localizes defects of varying sizes and combines image-level and feature-level anomaly scores to mitigate the impact of non-uniform distribution and noise. Experimental results on the infrared image dataset for aluminum foil sealing demonstrate high accuracy in anomaly detection and localization. Furthermore, on the industrial MVTec AD dataset, our TMIRN exhibits superior generalization and detection compared to state-of-the-art reconstruction networks.

Item Type:	Article
Date Type:	Published Online
Status:	Published
Schools:	Engineering
Publisher:	Institute of Electrical and Electronics Engineers
ISSN:	0018-9456
Date of First Compliant Deposit:	7 October 2024
Date of Acceptance:	28 September 2024
Last Modified:	17 Dec 2024 13:30
URI:	https://orca.cardiff.ac.uk/id/eprint/172655

Actions (repository staff only)

Edit Item