Wood Classification of Japanese Fagaceae using Partial Sample Area and Convolutional Neural Networks

Wood identification is regularly performed by observing the wood anatomy, such as colour, texture, fibre direction, and other characteristics. The manual process, however, could be time consuming, especially when work required at high quantity. Considering this condition, a convolutional neural networks (CNN)-based program applied to improve image classification results. research focuses on algorithm accuracy efficiency in dealing with dataset limitations. For this, it proposed do sample selection process or only take small portion of existing image. Still, can expected represent overall picture maintain generalisation capabilities CNN method stages. experiments yielded an incredible F1 score average up 93.4% for medium area sizes (200 × 200 pixels) each architecture (VGG16, ResNet50, MobileNet, DenseNet121, Xception based). Whereas DenseNet121-based was found best maintaining its model size (100, 200, 300 pixels). experimental results showed that accurate reliable solution.