Exploring Deep Learning Technique for Detection of Sterility Mosaic Disease in Pigeon Pea

Abstract

Pigeon pea is an essential legume crop in the semi-arid tropics and subtropics of Asia and Africa. Following chickpea, it ranks as the second most important pulse crop. Sterility Mosaic Disease (SMD) poses a significant threat to pigeon pea cultivation in the Indian subcontinent. This disease occurs daily and, under favorable conditions, can spread rapidly, leading to epidemics and causing substantial losses in pigeon pea production. Artificial intelligence techniques, specifically visual detection through the use of pre-trained Convolutional Neural Network (CNN) architectures such as VGG16, can aid in managing and mitigating the impact of sterility mosaic disease. Real-time and early quantification of the disease can play a crucial role in disease management and assist farmers in making informed decisions. Accurate and convenient disease detection in plants can enable the development of timely treatment methods and significantly reduce economic losses. In the case of Pigeon pea, CNN architectures pre-trained with VGG16 were utilized to train classifiers using a dataset comprising infected and healthy leaves collected from actual field experiments. Among the pre-trained architectures tested, the experimental results demonstrated an average accuracy of 82% in estimating sterility mosaic disease in Pigeon pea crop.