A Developed Deep Learning-based Approach for Microorganism Classification

Infections from microorganisms have become a significant threat to the growth and even life of the world in the past few decades. Most microorganisms pose a low-level danger to human life because they can spread through food, water, or the air. Every day, under normal conditions, this can lead to thousands of deaths worldwide. So, quickly and accurately classifying microorganisms is very important for things like clinical diagnosis, food safety, soil research, public health, and so on. By running processes directly on microscopic images, deep learning-based classification methods can correctly speed up identifying microorganisms. In this study, a developed approach for classifying microorganisms is suggested. In addition to the recommended way, five modern deep-learning approaches (Densenet201, Mobilenetv2, Resnet152v2, Inceptionresnetv2, and Nasnetlarge) are trained and tested on the collected microorganism dataset with and without data augmentation. The suggested approach worked better, with a F1-Score of 98%, a Precision of 98%, a Recall of 97%, and an Accuracy of 98.9473%.