ResNet101 Model Performance Enhancement in Classifying Rice Diseases with Leaf Images

Indonesia is the fourth biggest rice producer in Asia with its production accounting for 35.4 million metric tons yearly. This figure can increase unless crop failure resolved. Identifying diseases, however, may serve as an approach to minimizing risk of failure. The classification detect diseases was previously researched using ResNet101 method 100% accuracy. Despite this perfect accuracy, does not come without issue, where prediction yet optimal each label and loss results which are regarded too high due overfitting. Departing from research aims improve model by reducing layer complexity comparing two layers structures model, different data, model. performance resulting could be enhanced structuring simple architectural layers. small quantity dataset, yield accuracy a value 2.91%. experienced 2.7% at it accurately classify type according leaf images on label. problem solved that able disease even amount data utilizing appropriate arrangement requirements. In addition, overfitting occurred previous also resolved properly. matter proves correlation between very influential.