Rice is one of the important food crops for humans. The yield of this crop is affected by several factors, one of them is climate change. The irregularities of climate can cause the crops were infected by various diseases. By knowing the diseases through its unusual characteristics, fast remedial actions could be expected. The purpose of this essay is to find the best artificial neural network (ANN) architecture in classifying brown spot and leaf smut diseases on rice crops, two common types of diseases attack rice plant, based on the leaf images. Two features of rice leaf images i.e. shape and color were extracted to classify both diseases. The result of this research demonstrated the best ANN architecture to classify the diseases is the architecture with 4 neurons in the input layer, 5 neurons in one single hidden layer, and one neuron in the output layer. This architecture has accuracy as much as 66.3 percent.

No abstract