Abstract:
Coffee is grown in more than 50 countries in the world. And it is consumed worldwide in day-today life. In Rwanda it holds a unique position in the economy by making approximately 27 percent of total export revenue. Thousands relay on it as a livelihood. But its market share has not yet been fully developed because of several factors such as diseases, pests, insects, and limited use of advanced technologies. Coffee leaf rust (CLR) a coffee disease caused by a fungus called Hemileia Vastatrix is the most devastating one in Rwanda. It causes up to 50% leaf loss and up to 70% berry loss. This research is intended to develop an IoT and Machine learning based disease detection mechanism to monitor coffee leaf rust at early stage. A Pi camera sensor i are deployed to collect real-time data, a raspberry pi is configured to send data to google cloud platform and firebase for real-time data storage, analysis, visualization, and a web-based application is developed using FastAPI for user to access. A deep learning model is trained using ResNext algorithm, which performs 91% accuracy and it’s deployed in google cloud platform. The methodology used in this study is excremental where prototype is developed to collect data and secondary data from existing studies. The expected results from the study are a customized dashboard showing real time values of variables collected using prototype developed in a graph and in application.
