Wood species identification with transfer learning Available to Purchase

Wood or wood is a natural product of tree processing. Nowadays, the use of wood is decreasing, but some industries still use wood to manufacture their products to maintain product features. In some small industries, there are still errors in the use of wood species in production due to the lack of knowledge of industry players in identifying wood species. This study aims to develop a classification system for identifying wood species by applying transfer learning using a pre-trained model with InceptionV3 and MobileNetV2 architecture as its feature extraction layer. This research’s benefits are making it easier for woodworkers to identify wood types from wood images and speeding up the process of classifying wood types based on wood images. The dataset is 10624 wood images consisting of Agathis, Meranti, and Perupuk. The model used in this study was trained using 7659 images and validated using 1911 images. Performance testing of the classification model using the confusion matrix method was carried out on 1063 wood images. The results showed that the model with the best performance is mobileNetV2 with fine-tuning and 96% accuracy test results.