Pengenalan Daun Tanaman Obat Menggunakan Jaringan Syaraf Tiruan Backpropagation

Abstract

Indonesia is a country with a variety of biodiversity. One of the rich types of flora or plants is medicinal plants. Not all types of medicinal plants can be remembered by the community because people have limited memory. In addition, the many types of medicinal plants make an error in the process of introduction of medicinal plant types. This research processes leaf images using image processing. The data used in this study 189 data consisting of 7 types of medicinal plants. Feature extraction used was 21 features which included shape, texture and color. The result of feature extraction will be identified using backpropagation neural network. Classification experiments with backpropagation produce an optimal accuracy of 91%. These results are generated using data normalization and 3 fold. The architecture used is input 21 features, 40 neurons in the hidden layer 1 and 15 neurons in the hidden layer 2. The results are obtained by the trainscg function, the activation function tansig.