Cinnamon Plant Disease Identification System Using Machine Learning and Deep Learning

Abstract

Cinnamon is a major agricultural export product in Sri Lanka. There is a high demand for Sri Lankan Cinnamon on the global market. Therefore, more than 9000 farmers cultivate cinnamon. Unfortunately, many diseases affect the growth and quality of the cinnamon plant. There is a lack of experts who can identify diseases affecting cinnamon plants. Early disease identification is very important to cure diseases in cinnamon. If not, the growth and quality of the cinnamon plant become low. This becomes a huge problem in the Cinnamon industry. To solve above mentioned problem, the author aims to develop a system using machine learning and deep learning. By uploading an image of a disease-infected area, farmers can identify diseases accurately without the help of an expert. With the use of the system, farmers can see the disease-infected percentage, visual representation of the disease and treatment recommendations. By using this system farmers can easily identify diseases in real time without paying any cost for expertise. The system's effectiveness was carefully evaluated using a variety of data science criteria, including accuracy, precision, recall, and F1-score, to ensure a complete performance assessment. During the testing phase, the model showed a high accuracy rate of 87.7% and a precision of 87.8%, indicating the system's capacity to properly detect disease-infected areas. The 27.3% recall rate indicates that the system is very capable of recognizing all relevant cases of the disease within the dataset. Furthermore, an F1-score of 41% shows the model's balanced accuracy and recall, demonstrating its resilience. These indicators demonstrate the system's ability to deliver trustworthy disease detection, enabling farmers with quick and precise disease management tools.