Rice Disease Detection Using Faster R-CNN Approach

It has never been easy to identify plant diseases accurately and quickly. A significant amount of food grains is lost by farmers each year as a result of the absence of automated tools that can precisely detect plant diseases. At the moment, depending just on eye observations requires a great level of experience and cannot be assured to be accurate. In order to solve these issues, we suggested a way to recognise and categorise rice plant illnesses. A publicly accessible benchmark dataset including 1800 photos was enhanced to include 12,600 images. Four classes are included in the dataset: Leaf Blast, Brown Spot, Hispa, and Healthy Leaves. We used the Faster R-CNN approach for disease diagnosis and classification, adjusting parameters to increase precision. Our approach can help with early diagnosis with high precision, with accuracy of 98.20%, 99.15%, 97.5%, and 99.5% for Brown Spot, Hispa, Leaf Blast, and healthy leaf identification, respectively. Our approach can be enhanced for practical agricultural use and help farmers avoid crop damage and disease.