Institutional Repository of Vilnius University
Few-Shot learning for image-based anomaly detection /
Abstract
dc:descriptionTraditional quality control methods used in the manufacturing industry are often expensive and prone to human error. In the context of Industry 4.0, neural network based technologies are increasingly being introduced to automate and improve the process of defective product detection. However, training neural network models typically requires large labeled datasets, which are often difficult to obtain in manufacturing environments due to the rarity of anomalous samples. The aim of this work is to develop a Few-Shot Learning (FSL) extension for the PatchCore model by introducing the proposed DualPatchCore architecture, which incorporates an additional memory bank for storing anomalous feature embeddings. The proposed model is implemented, trained, and evaluated against a collection of benchmark models, including CFA, EfficientAD, and PatchCore.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Pilipavičius, Domantas,
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:ELABAETD308338368&prefLang=en_US