Abstract The classification of mushrooms is an essential endeavor for ensuring food safety, aiding biodiversity research, and averting accidental poisoning. Conventional methods that rely on visual and physical traits necessitate expert knowledge and are susceptible to human error. To overcome these challenges, this project introduces a machine learning-based system designed for the automated classification of mushrooms as either edible or poisonous, utilizing structured datasets and supervised learning algorithms. The system utilizes various mushroom characteristics—such as cap shape, gill structure, odor, and habitat—to train and assess classification models, including Decision Trees, Random Forests, Support Vector Machines (SVM), and Neural Networks. Techniques for data preprocessing, feature extraction, and model optimization are applied to improve the accuracy of the system, achieving high reliability and generalization across different species. A user-friendly interface facilitates real-time predictions along with confidence scores, delivering informative and accessible results to users such as foragers, researchers, and food safety professionals.
kr et al. (Sun,) studied this question.