An Internet of Things IOT Based Automated Irrigation and Pomegranate Plant Disease Detection System Using Deep Learning Algorithms

Abstract

Agriculture is a paramount factor that contributes in eradicating extreme poverty and moreover it accounts in enhancing the Gross Domestic Product (GDP) of a country. Producing a quality crop in large quantities is a big challenge to farmers as there are many factors impacting it, like soil, moisture content, erratic weather condition and disease of plants due to fungal, viral infection and pests. So there is need of precision agriculture with the use of proper technology that makes the entire process automatic and control parameters like moisture content to keep the crop healthy. Pomegranate cultivation in India, especially in Maharashtra is one of the major sources of revenue generation for farmers. There is very less work done in regards to pomegranate plant/cultivation, hence farmers are facing crop losses due to erratic weather condition and fruit diseases. Existing research carried out in plant leaf/fruit disease prediction uses public dataset, moreover they focus only on leaf disease or fruit disease or moisture content estimation. There is need of holistic system that automatically monitors and regulates moisture content, alert temperature variations and predict early detection of pomegranate plant. This research work carried out integrates automatic irrigation system, leaf disease detection and fruit disease detection. Moreover it guides the farmer with relevant pesticides as prescription for disease detected /predicted. It incorporates Internet of Things to automate irrigation system for pomegranate plant and leaf/fruit disease detection. The research work carried out, uses one of the commonly used deep learning architecture called Convolutional Neural Networks (CNN) for developing the model. The CNN algorithms are one of the best developing algorithms used for image processing and classification. The CNN model is trained with the real time images of healthy and diseased leaves similarly for pomegranate fruit disease detection, the CNN model is trained with public pomegranate fruit dataset. A total of 6