| Abstract [eng] |
When deep learning methods started to show state-of-the-art performance for solving complex object recognition problems, researchers begun attempting to apply these methods to solve other problems too. One of those problems is image super-resolution. In this work the problem of image super-resolution was analyzed, artificial neural networks and their use to solve this problem were reviewed. Super-resolution convolutional neural network was implemented, various modifications were made and their influence on the final result was examined. Problem of image cropping when using this method was analyzed and a solution was proposed. Better results were reached with the help of implemented modifications. |
