SwinMTL: A Shared Architecture for Simultaneous Depth Estimation and Semantic Segmentation from Monocular Camera Images | IEEE Conference Publication | IEEE Xplore