A Comparative Analysis of Machine Learning Algorithms for Bok Choy Leaf Disease Identification in Smart Aquaponics

Urban gardening innovation is driven by food scarcity. Smart aquaponics is one of the strategies utilized in urban farming. However, when plants and crops are infected, agricultural output is impacted. Farmers study plants visually for disease detection and identification. It is time-consuming, expensive, and inaccurate. A smart aquaponics system must be monitored, regulated, and automated to succeed. Using vision systems and machine learning algorithms to enhance farming is efficient. To do this, a comparison of three machine learning estimators was conducted: the Support Vector Machine (SVM), the Random Forest (RF), and the K-Nearest Neighbor (KNN). Each algorithm was modeled on Bok Choy leaves from a smart aquaponics system. Each model has been optimized to maximize hold-out validations. The results indicated that KNN with tuned hyperparameters was the most successful model for the dataset, with a mean accuracy of 98.21% for hold-out validation and 98.25% for classification.