Ansah, Michael Kwame Atansa2022-12-222022-12-222021-05https://hdl.handle.net/20.500.11988/798Capstone Project submitted to the Department of Engineering, Ashesi University in partial fulfillment of the requirements for the award of Bachelor of Science degree in Computer Engineering, May 2021Tomato production in Sub-Saharan Africa can be plagued by a fungal disease known as leaf blight. The tomato blight disease can have devastating effects on smallholder farmers if measures are not taken early. Currently, the most common method of leaf disease diagnosis is an estimation by human judgment. This method is very error-prone and can be quite tedious given the volume of plants on a typical tomato farm. This research investigates the creation of a faster disease detection method using image processing and machine learning. Similar solutions in the past that employ these technologies typically have to connect to an offsite server to perform all processing and diagnosis. This project explores the deployment of compressed, quantized models on small edge devices which can produce accurate results at a significantly low cost. The tomato blight disease is characterized by brown leaf spots. The system could classify a diseased leaf with brown spots from a healthy leaf with an accuracy of 90% ± 12.32.enroboticsagricultural surveillance devicesCapstone project