A Semi-Supervised Approach to Monocular Depth Estimation, Depth Refinement, and Semantic Segmentation of Driving Scenes using a Siamese Triple Decoder Architecture. INF. 2020;44(4). doi:10.31449/inf.v44i4.3018