Wood Type Classification Using Deep Learning Approach

Abstract

This research investigates the efficacy of various deep learning models for wood type classification, employing a dataset comprising 2011 images across five distinct wood classes. Six state-of-the-art architectures, including ResNet-50, DenseNet121, MobileNetV2, VGG16, VGG19, and Inception-V3, were implemented and evaluated. Training and model accuracy were assessed to determine each model's performance. Results reveal that ResNet-50 achieved the highest accuracy, reaching 98%, closely followed by DenseNet121 and MobileNetV2, both achieving 97%. VGG16 and Inception-V3 attained accuracies of 94% and 92%, respectively, while VGG19 achieved 92%. These findings underscore the potential of deep learning approaches in accurately classifying wood types, with ResNet-50 demonstrating superior performance among the models tested. This study contributes valuable insights into utilizing deep learning methodologies for wood type classification, offering practical implications for industries reliant on wood classification for various applications.