Deep Learning-Based Mushroom Classification: Differentiating Edible and Poisonous Species Using Convolutional Neural Networks

Damage caused by mushroom poisoning has significantly risen over the years. It is hard to determine the difference between edible and poisonous ones, for they have similar physical features. However, this study has introduced a system using image processing algorithms utilizing Convolutional Neural Network (CNN) to extract important characteristics from mushrooms to identify edible and poisonous varieties. Samples were tested wherein images of mushrooms were aided with Python, specifically Jupyter and Spyder Notebook, to create different Deep Learning Models and identify the most accurate one using various metrics such as Accuracy, AUC, Min Squared error, Precision and Recall. The developed system was able to have an accuracy rate of 99.08% indicating that the model can effectively classify the edible and poisonous mushrooms.