DeepThaiMush: A Deep Learning Approach to Classify Poisonous and Edible Mushrooms using YOLOv8

In the field of mushroom distinction, it is crucial to identify the differences between edible and poisonous varieties. Since many mushrooms can look alike and mistakes can be dangerous, we need fast and trustworthy ways to classify them. This study aims to utilize the capabilities of YOLOv8, an innovative deep learning approach, for classification for mushrooms found in Thailand as edible or poisonous. Specifically developed for high-speed object detection, YOLOv8’s unique architecture sets it apart from conventional methods. When applied to mushrooms found in Thailand, YOLOv8’s unique methodology was pivotal in achieving an impressive precision of 99.32%, a perfect recall rate of 100.00%, and an F1-Score of 99.66%. When we compared YOLOv8 against other methods, our proposed approach performed on par with GoogLeNet (F1-Score of 99.67%) and ResNet-50 (F1- Score of 99.66%), and surpassed the Improved AlexNet which had an F1-Score of 99.09%. These findings prove that YOLOv8 is an excellent choice for real-time mushroom identification.