Article: Few-shot learning in industrial applications
Alternative Title
Authors
Molek, Vojtech
Alijani, Zahra
Editor
Abstract
This paper reports on the empirical performance of few-shot learning (FSL) for visual defect classification using confidential industrial datasets. We evaluate 16 combinations of four backbone models (Perception Encoder, DINOv2, DINOv3, ConvNeXt-v2) and four FSL classifiers (Prototypical Networks, Neighborhood Component Analysis, Relation Networks, Linear Adapter). The evaluation covers three conditions: a baseline comparison, deterministic support set augmentation, and a learnable attention preprocessor. Results demonstrate that support set augmentation is a highly effective strategy, improving performance in nearly all configurations. Furthermore, the DINOv2 and ConvNeXt-V2-T backbones emerged as top performers, achieving the most competitive and highest-accuracy results, respectively. These findings suggest that for industrial FSL applications, combining a strong, pre-trained backbone with a simple augmentation strategy is a practical approach for building data-efficient classification systems.
Description
Subject Headings
fuzzy systémy, matematika, informatika
Keywords
fuzzy systems, mathematics, informatics
ISBN
ISSN
DOI
10.15452/978-80-7599-515-5.2026.14
License
CC BY 4.0
Citation
MOLEK, Vojtech and ALIJANI, Zahra, 2026. Few-shot learning in industrial applications. Online. In: STUPŇANOVÁ, Andrea; DYBA, Martin and PAVLISKA, Viktor (eds.). Proceedings of The Eighteenth International Conference on Fuzzy Set Theory and Applications. Ostrava: University of Ostrava, p. 74-77. ISBN 978-80-7599-515-5. Available at: https://doi.org/10.15452/978-80-7599-515-5.2026.14.