Integrating Attention Mechanism and Optimizers for Enhanced Wood Surface Analysis and classification with a Novel Dataset
Keywords:
Deep Learning, ResNet18, Bottleneck Attention Mechanism, Wood Surface Analysis, Texture classification.Abstract
The wood industry faces significant challenges due substantial variability in raw materials and the complexity of manufacturing processes, which result in numerous visible structural defects. Manual quality control, reliant on trained specialists, is often tedious, biased, and less effective. Automated vision-based systems have been proposed as a solution, achieving higher recognition rates than human experts. However, the field suffers from a lack of large-scale, authentic datasets that encompass both normal and defective wood surface images.
References
Abdullah, N. D., Hashim, U. R. A., Ahmad, S., & Salahuddin, L. (2020). Analysis of texture features for wood defect classification. Bulletin of Electrical Engineering and Informatics, 9(1), 121-128. DOI: https://doi.org/10.11591/eei.v9i1.1553
Kodytek, P., Bodzas, A., & Bilik, P. (2021). A large-scale image dataset of wood surface defects for automated vision-based quality control processes. F1000Research, 10. doi: 10.12688/f1000research.52903.2
