Indonesia as an agrarian country with the largest population uses rice as a staple food is depending on rice production. The lack of quantity and quality of production that often occurs is caused by disease attacks on plants that are detected too late. This is due to the lack of agricultural extension workers who help farmers in dealing with plant diseases. This study conducted an experiment on rice plant diseases based on leaf imagery using a dataset that has four classifications of leaf conditions of rice plants, namely healthy, brown spot, hispa, and leaf blast. The results obtained are quite good, namely, the accuracy value of the training data is 88.4% and 82.99% in data testing using the Densely Connected Convolutional Networks (DenseNet)-201 architecture as. From the results of the key research, DenseNet201 is quite suitable to be used to carry out diseases in rice plants so that the types of diseases that attack can be identified and given early. Thus food security can be maintained and not cause losses due to crop failures that harm farmers.
Copyrights © 2022