Application of Faster R-CNN Deep Learning Method for Rice Plant Disease Detection

Abstract

Plant diseases, particularly in staple crops like rice, significantly affect the stability of rice production in Indonesia. Crop failure caused by rice plant diseases present a critical challenge for farmers. Early diagnosis is crucial for preventing and managing rice diseases, as it facilitates more effective preventive measures, reduces yield losses, and boosts overall agricultural production. This study aims to apply the Faster Region Convolutional Neural Network (Faster R-CNN), a deep learning approach, to detect rice plant diseases. The Grid Search method was employed as a hyperparameter tuning technique to identify the optimal parameter combination for enhancing algorithm performance. Experimental results demonstrate the model's performance, achieving an accuracy rate of 88%, recall and precision of 100%, and an F1 Score of 93%. These findings indicate that the Faster R-CNN method effectively recognizes and classifies rice plant diseases with a high degree of accuracy.

Keywords

• faster r-cnn
• grid search
• deep learning
• hyperparameter tuning
• rice plant disease detection