Rozpoznání výrazu ve tváři pomocí konvolučních neuronových sítí

Abstract

This thesis is dealing with face expression recognition. The goal was to create a classifier based on convolutional neural networks, which would be able to classify emotions on human face. In this thesis are described methods which can lead to successfull solution and compiled classifier was based on these methods. This classifier achieved a test accuracy of 72.9% on FER2013 dataset. In conclusion are summarized gained knowledge, achieved results and suggestions for improvement.