SOSD-Net: Joint semantic object segmentation and depth estimation from monocular images

Depth estimation and semantic segmentation play essential roles in scene understanding. The state-of-the-art methods employ multi-task learning to simultaneously learn models for these two tasks at the pixel-wise level. They usually focus on sharing common features or stitching feature maps from corresponding branches. However, lack in-depth consideration correlation of geometric cues parsing. In this paper, we first introduce concept objectness exploit relationship through an analysis imaging process, then propose a Semantic Object Segmentation Estimation Network (SOSD-Net) based assumption. To best our knowledge, SOSD-Net is network that exploits geometry constraint simultaneous monocular depth segmentation. addition, considering mutual implicit between tasks, iterative idea expectation–maximization algorithm train proposed more effectively. Extensive experimental results Cityscapes NYU v2 dataset are presented demonstrate superior performance approach.