Pneumonia is a contagious disease of the lower respiratory tract that occurs when bacteria, viruses or fungi attack the lungs and lead to an inflammatory response that protects the human body from infections. In the worst case, the result of pneumonia can be the death of a person. It is not uncommon for someone to get pneumonia, especially in developing countries where large numbers of people face energy poverty and rely on polluting forms of energy. This paper deals with the application of convolutional neural networks for the detection of pneumonia from magnetic resonance imaging images. The theoretical part of the paper explains artificial neural networks, techniques used to improve the operation of neural networks, convolutional neural networks, GradCAM techniques and the obtained results. In the practical part of the paper, different architectures of convolutional neural networks as well as techniques for improving the operation of neural networks were used to classify magnetic resonance images. The GradCAM technique was used to visualize what the convolutional neural network pays attention to when classifying a particular image. A web application was also created through which the diagnosis of pneumonia can be made on the basis of the uploaded image. The programming code is written in the Python programming language, and the Tensorflow programming framework is also used.