Paddy Crop Disease Detection Using Deep Learning

Abstract

Agriculture plays a crucial role in human life, with approximately 60% of the population directly or indirectly involved in agricultural activities. Paddy is one of the essential food crops globally and is particularly significant in the Asian subcontinent. As a result of excessive use of chemicals and unpredictable weather patterns, there has been a significant increase in crop diseases. Sometimes an expert may be unavailable to identify the disease. Due to mistaken conclusions of experts, there is an unnecessary use of pesticides which will affect the yield badly, hence, it is essential to know which disease has affected the Paddy crop. Early detection of these diseases is essential to minimize the losses . To address this issue, Deep Learning models, including Artificial Neural Network (ANN), Convolutional Neural Network (CNN), and ResNet101, were employed to detect three types of paddy crop diseases including Leaf Blast (LB), Brown Spot (BS), and Hispa along with the healthy category. The dataset consisted of 6,061 images of three types of disease affected and healthy paddy crops, collected from Paddy Doctor Website and IRRI. ANN Model achieved an accuracy of 66.1%, CNN Model 94.3% and ResNet101 Model 98.2%.