Institutional Repository of Vilnius University
Anomaly detection in industrial manufacturing using deep neural networks /
Abstract
dc:descriptionAnomaly detection in industrial manufacturing has evolved from initial human inspection to advanced deep-learning methods. Initially, workers monitored assembly lines to spot defects. Nowadays, various deep-learning models can efficiently detect anomalies in product images, enhancing the ability to identify and prevent faulty products during production. This study aims to create a new deep-learning model for industrial anomaly detection by integrating the Asymmetric Student-Teacher model into the architecture of the EfficientAD model. The developed network is evaluated on "MvTec" software's datasets, which consist of images with industrial manufacturing defects. The collected metrics are compared with existing deep neural networks for industrial anomaly detection to evaluate categories upon which the modified model showed better or worse results. The developed model shows better results than several state-of-the-art models in some categories and improves the original EfficientAD model on certain statistics for image classification and segmentation tasks.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Dulevičius, Dominykas,
Subjects
dc:subject × 1Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- eng
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD210647464&prefLang=en_US