Machine Learning Algorithm based Disease Detection in Tomato with Automated Image Telemetry for Vertical Farming

This paper is highlighting an outline of disease detection in tomato using computer vision and machine learning algorithms. Readily available hardware is used to build a system where a camera mounted system can detect and identify spot disease in tomatoes in real-time. As an initial prototype only spot disease can be detected. The complete development can be divided into two parts. The first part is the software and algorithm which aimed to detect and identify disease in crops and generate a report for the user. It is successful in building the algorithm and GUI (graphical user interface) for the user which can detect spot disease in tomatoes. Using the Viola-Jones algorithm and Haar like feature extraction method for the machine learning process in MATLAB, an XML (an image trained file) file for spot disease in tomatoes is designed using 377 images of infected tomatoes. The second part is the hardware implementation which consists of a simple robot rig that carries the camera and the system scans the tomatoes for the disease. For the vast majority of the time, spot detection is accurate. Many other diseases which exist for the animal, human and crops can easily be added to the system. In terms of reliability, the system is a success with acceptable false positives.