Mushroom Classification Using Deep Learning

Abstract

As a useful and diverse part of ecosystems, mushrooms are important in many areas, such as the gastronomic, medicinal, and environmental. To classify 6000 photos of mushrooms from various species, this study uses deep learning techniques. The principal aim is to create a reliable and precise model that can distinguish between various types of mushrooms, thereby advancing the field of mycology and facilitating applications linked to mushrooms. The collection of photos includes pictures of several types of mushrooms, such as Reishi, oyster, pink oyster, golden oyster, button, and blue oyster mushrooms, among others. A complete and well-balanced dataset for training and assessment is provided by the 1000 photos that each class represents. Resizing, contrast stretching, and gamma correction are some of the augmentation techniques used to improve the dataset by creating variances in the photos. Cuttingedge deep learning models like ResNet50, MobileNetV2, VGG16, and VGG19 are used and optimized for the classification task. Using transfer learning, the classification performance for mushroom photos is optimized by utilizing the information obtained from pre-trained models on big datasets. The project intends to address particular issues related to the variety of mushroom forms, colors, and attributes in addition to achieving high classification accuracy. The goal of the research is to determine the best architecture for tasks involving the classification of mushrooms through a comprehensive process of experimentation and comparison analysis of several models. To further emphasize the model's practical relevance, the proposed work also includes an evaluation of the model using real-world mushroom datasets. Metrics including accuracy, precision, recall, and F1 score will be used to analyze the classification findings. Potential uses of the proposed model in forestry, environmental monitoring, and agriculture are also covered in the study