Automated Mushroom Classification System using Machine Learning

Abstract

The mushroom farming industry is on the rise and there is a growing demand for high-quality, sustainable production methods. However, mushroom cultivation can be challenging for growers, as it requires careful environmental control and can be labour-intensive. This paper presents the results and analysis of a mushroom classification system, designed to accurately classify two types of mushrooms: Shiitake and oyster mushrooms. The system utilizes advanced machine learning algorithms and an extensive dataset to achieve accurate classification results. Technology replaces human inspection with computer vision, providing farmers with a quick and precise tool for classifying various mushroom species. The augmented and labelled mushroom dataset is divided into three distinct subsets: validation, testing and training. In this study, 80% of the data is allocated for training, while 20% is reserved for testing. The outcomes have proven the efficacy of You Only Look Once version 8 (YOLOv8) model, with an astounding accuracy of over 90%. This study suggests that the YOLOv8 model has the potential to be an accurate method for distinguishing different varieties of mushrooms. Farmers can significantly reduce the time and effort they previously expended on manual inspections by utilizing this approach.