An Efficient Deep Learning Based Approach for Toxic and Edible Mushroom Classification

The mushroom industry has grown tremendously worldwide, and it has greatly contributed to recent gains in the GDP. However, there are significant health risks associated with eating toxic or poisonous mushrooms due to this growing demand. Advancements in Artificial Intelligence and Machine Learning algorithms have significantly improved the ability to classify edible and toxic mushrooms. In this study, we developed a customized deep learning model that integrates the feature extraction capabilities of EfficientNet-B0 with a Support Vector Machine (SVM) for classification. Through comparison with other deep learning architectures, performance was confirmed, proving the suggested model’s greater effectiveness. Our unique model outperformed current and conventional deep learning techniques by achieving an astounding 99.5% accuracy rate.