ResNet-50 to Classify the Types of Indonesian Local Coffee Beans Muhammad Aji Alrasyid (a), Bagas Rohmatulloh (b), Retno Damayanti (b), Dimas Firmanda Al Riza (b), Mochamad Bagus Hermanto (b), Sandra (b), Yusuf Hendrawan (b*)
a) Department of Statistics, Faculty of Mathematics and Science, Universitas Brawijaya, Malang, Indonesia
b) Department of Agricultural Engineering, Faculty of Agricultural Technology, Universitas Brawijaya, Malang, Indonesia
*yusuf_h[at]ub.ac.id
Abstract
A technology is needed to distinguish the types of Indonesian local coffee beans to prevent product counterfeiting. One method for detecting agricultural product varieties is computer vision. This study aimed to classify three types of Indonesian Arabica coffee beans i.e. Gayo Aceh, Kintamani Bali, and Toraja Tongkonan using computer vision. The classification method used was the ResNet-50 convolutional neural network with sensitivity analysis using several variations of the optimizer such as SGDm, Adam, and RMSProp, as well as the learning rate of 0.00005 and 0.0001. Each type of coffee used 500 data for training and validation with the distribution of 70% training and 30% validation. ResNet-50 model using the RMSProp optimizer at a learning rate value of 0.00005 can be recommended for several reasons i.e. the training process was more stable with little fluctuations, had the highest value of validation accuracy of 100%, and highest testing accuracy of 99.6%.
