EARLY WARNING SYSTEM FOR COFFEE LEAF RUST DISEASE ON ROBUSTA PLANT

• Main Author: Salman Alfarizi, Muhamad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/75408
• Institution: Institut Teknologi Bandung

In this study, an IoT-based early warning system for leaf rust disease attacks on robusta coffee plants was designed and implemented. This system monitors five variables, namely temperature, humidity, light intensity, rainfall, and the current image of the plant. This IoT-based monitoring system is designed to send data on all variables in robusta coffee plantations in Pamulihan District via MQTT to a server at the Bandung Institute of Technology. Temperature and humidity measurements were made using a BME280 sensor, while light intensity measurements were made with a BH1750FVI sensor. Rainfall values were extracted using web scraping techniques through 'the weather channel' website. The image obtained through the camera is sent to the server. On the server, object detection takes place using the YOLOv8 algorithm. The resulting YOLOv8 model has a mAP50-95 value of 0.708. The five variables obtained were grouped using the k-means clustering algorithm. The elbow method is iteratively used to find the correlation between environmental condition data and images that detect leaf rust disease. The group indicates the condition of plants susceptible and resistant to leaf rust disease. The inertia value obtained in the 2 groups is 8.61. Based on the results of the elbow analysis, the system can send early warnings via email about the occurrence of leaf rust disease attacks for the next 6 days.