Research study of image classification algorithms based on Convolutional Neural Networks

The beginning of the paper introduces the idea of neural networks by presenting the neuron models and artificial neural networks. Next, the flow of operations used in the development of convolutional neural networks is presented, as well as an overview of popular architectures used in model development. After introducing the theoretical issues, a project using the EfficientNet architecture to recognize medical images taken by MRI is presented. The project describes: the selection and parameters of the MRI image database representing different types of brain tumors, the process of implementation and appropriate modification of the EfficientNet architecture used to classify images, the presentation of the results of learning the convolutional neural network and the images that highlight on the basis of which area the convolutional neural network classifies images.