Classification of Aspergillus fungi species using convolutional neural networks

Mold fungi are among the most common types of fungi found worldwide. These fungi are commonly found in soil, plants, rotten fruit, vegetables and foods. Mold fungi cause damage to agricultural plants and serious dangers in terms of human health. Especially Aspergillus species are among the species that are dangerous for human health. There are many different types of Aspergillus fungi in nature. In this study, species classification was made on Aspergillus Flavus and Aspergillus Fumigatus fungi. The Convolutional Neural Networks (CNN) model, which is one of the widely used Deep Learning algorithms was used for the classification of Aspergillus species. In the literature, there are many studies with successful result in image recognition and classification with convolutional neural networks. Convolutional neural netvorks are widely used in the identification and classification of different kinds of plants, bacteria, living things and objects. In this study, a data set containing a total of 1050 images was used. The dataset was prepared by Prof. Dr. Yasemin Öz, a faculty member of Eskişehir Osmangazi University, Faculty of Medicine. A convolutional neural network model was created and various experiments were carried out on this model. Experiments were performed on both color and grayscale images. As a result of experiments on grayscale images, the training accuracy was 0.9777 and the test accuracy was 0.9095. Then the result obtained by applying the five fold cross-validation method were compared.