During this project, state-of-the-art deep learning models have been used to estimate depth maps from a monocular RGB image applying a teacher-student learning approach. This paradigm has been used in order to distillate the knowledge of high capacity deep neural networks into shallower ones to make inference faster for real-time applications. Some successful applications of this technique can be found both at natural language and computer vision applications.

This work will focus on studying different deep learning architectures for obtaining depth information from monocular RGB images.