Detection of Paddy Crops Diseases and Early Diagnosis Using Faster Regional Convolutional Neural Networks
K Anandhan., Ajay Shanker Singh
Abstract
Nowadays the grain food demand is increasing due to population increase. Rice is one of the essential foods for half of the world's population. In India most of the farmer struggled to identify paddy leaf disease at a premature stage. The deep learning models were used to detect a rice blast disease such as Brown spot, Sheath blight, Blast and Leaf streak disease at early stage, so we can control disease spread over all plant to increase the rice production. The proposed system is applied to identify a various rice plant leaf images disease detection using mask R-CNN, Faster R-CNN algorithms. We have studied five different paddy diseases and applied in southern region (India) to improve the paddy quantity. We collected a paddy leaf images through a Sony Camera & Mobile device (Vivo V9) and conducted experiments to train the model using 1500 images. Our experimental results analysis mask R-CNN is best suitable for to detect and identify the various rice blast disease such as Blast-96%, Brown spot-95%, and Sheath blight-94.5%. The results implies capable of disease identification from real-time capture leaf images and improve the accuracy applying different areas to control the spread rice blast disease.