Implementasi Algoritma Convolutional Neural Network dalam Menentukan Kelayakan Kayu

Abstract

Wood eligibility is the most important factor in the furniture industry. However, currently there are still many producers who ignore the feasibility of wood so that it can affect production results and selling prices. With the development of technology such as digital image processing, the process of selecting feasible wood can be done without the need for human visuals. This research proposes to classify wood eligibility based on digital images of wood eligibility using the Deep Leraning method. Convolutional Neural Network (CNN) which is one type of Deep Learning algorithm is proposed as a method to analyze wood worthiness images. The dataset of wood worthiness images was obtained through observations made by researchers at CV Kanindotama. The dataset used in this study amounted to 105 wood images divided into 83 training data and 22 test data. The model built using the ResNet50V2 architecture gets the greatest accuracy of only 69.51% for training data and 62.5% for test data. While the model built using the MobileNetV2 architecture gets an accuracy of up to 98.29% for training data and 100% for test data. This proves that the MobileNetV2 architecture is better than ResNet50V2. In addition, it can be said that the CNN algorithm can be used to analyze the feasibility of wood well.