Konvolucijske neuronske mreže za klasifikaciju objekata

This graduate thesis deals with convolutional neural networks for object classifiaction problem. Part one of this graduate thesis deals with artificial neural networks, their history, their architecture, activation functions, process of learning and backpropagation algorithm. Part two of this graduate thesis deals with convolutional neural networks, convolution, picture processing filters, architecture of convolutional neural networks, backpropagation algorithm and a few examples of convolutional neural networks. Part three of this graduate thesis deals with convolutional neural networks for object classification in a few steps: analysing HAM 10000 dataset, preprocessing of the data, building model for object classification and analysing results with future plans. The aim of this graduate thesis was to explain everything about convolutional neural networks and use them on very important problem of skin leasions classification.