Machine learning-based classification of mushrooms using a smartphone application

Authors: LEE, JAE - Purdue University; AIME, M - Purdue University; RAJWA, BARTEK - Purdue University; BAE, EUIWON - Purdue University

Submitted to: Applied Sciences
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/4/2022
Publication Date: 11/17/2022
Citation: Lee, J.J., Aime, M.C., Rajwa, B., Bae, E. 2022. Machine learning-based classification of mushrooms using a smartphone application. Applied Sciences. 12(22):11685. https://doi.org/10.3390/app122211685.
DOI: https://doi.org/10.3390/app122211685

Interpretive Summary: Consumption of harmful and poisonous mushrooms lead to numerous illnesses and deaths world-wide each year. This is in part due to the similarity between poisonous and edible mushrooms. Here we report a android-based cell phone application to help non-mycologists to determine if mushrooms are edible. The application allows users to obtain a probability score for the mushroom classification. Using real-life, field-based images that contained diverse backgrounds and objects, the application was able to provide sensitivity and specificity ranging from 89%-100%. Further refinement of the classification system will allow a robust application for mushroom identification, potentially saving lives.

Technical Abstract: Worldwide, a large number of cases of harmful mushroom exposure and consumption result in hallucinations, sickness, and death. One contributing factor is that certain poisonous mushrooms closely resemble their edible counterparts, making it difficult for general public collectors to distinguish one from the other. Purdue scientists propose a method to classify mushroom types from field collection images using a smartphone application based on a convolutional neural network. The application helps people without proper mycology background or training to distinguish poisonous mushrooms from edible ones with which they may be confused. Three case studies were shown to classify two-, three-, and five-class models by optimizing their training steps by cross-validation. An android app was developed by transferring the server-based trained model and allowing users to obtain probability scores for the correct genus classification. The experiments showed that this method could provide sensitivity and specificity of two-, three-, and five-class mushroom models ranging from 89% to 100% using an image from the field with diverse backgrounds and objects.