Leveraging Deep Learning for Accurate Classification of Wild Mushrooms: A Hybrid ResNet and Region-Based CNN Approach

Accurate mushroom classification is essential for applications ranging from ecological research to foraging safety. This study proposes a convolutional neural network (CNN) based approach for multi-class mushroom classification, aiming to address the unique challenges posed by diverse morphological features across mushroom species. Utilizing an expanded dataset of nine classes, this model was trained and evaluated with various metrics, including accuracy, F1score, ROC-AUC, and a detailed analysis of confusion matrices, to assess performance across class boundaries. Future improvements will focus on expanding the dataset to include more diverse samples, implementing classspecific training weights, and further refining CNN layers to enhance feature extraction. These advancements will address current limitations and improve the model’s generalizability and robustness, underscoring its potential as a robust tool for mushroom classification applications.