Smallholder farmer-centric integration of IoT and Chatbot for early Maize diseases detection and management in pre-visual symptoms phase

Abstract

Crop diseases are responsible for more than 50% of worldwide yield loss. Therefore, detecting crop diseases as soon as possible is important to limit both yield loss and costs involved in interrupting the disease cycles. Traditional disease detection sensors such as farmer eyes or crop imaging devices backed by Deep Neural Networks (DNN) models rely on visual symptoms which appear after the disease cycle has already undergone several asymptomatic phasing events such as inoculation, penetration, and so on. Recently, Internet of Things (IoT) sensors have been used to monitor other pre-visual abdominal signs generated by infected plants, like for instance emission of Volatile Organic Compound (VOC) or changes in soil nutrition consumption patterns. Considering the limited availability of extension officers to timely support farmers in early disease management, in this paper we propose an integrated framework that converges IoT sensing asymptomatic signs with natural language processing (NLP) chatbots to enable low literacy smallholder farmers of Maize crops in East-Africa to be completely autonomous in identifying and understanding their potential crop diseases prior to the apparition of visual symptoms.