Classification of Microscopic Images of Fungi Using Deep Learning Models

Abstract

The paper aims to identify the microscopic images of fungi Colletotrichum gloeosporioides and Cylindrocladium colhounii, which affect the leaves of Mango and Custard apple plant. The affected leaves are collected, cultivated and observed under high power 40X objective lens. The collected dataset is brand new of 78 images and feed to the two pre-trained deep learning models AlexNet and SqueezeNet. We propose an idea of identifying these fungi at microscopic level, where not much of work is done on fruit plants of Hyderabad and Karnataka (H8K) region of Karnataka, India. Using the CNN models, performance measures such as mean precision, mean recall, F1 score and accuracy are calculated and it is observed that AlexNet model gives a good recognition accuracy of 93.8% compared to the SqueezeNet model.