Deep learning for monocular depth estimation: A review

Depth estimation is a classic task in computer vision, which of great significance for many applications such as augmented reality, target tracking and autonomous driving. Traditional monocular depth methods are based on cues prediction with strict requirements, e.g. shape-from-focus/ defocus require low field the scenes images. Recently, large body deep learning have been proposed has shown promise handling traditional ill-posed problem. This paper aims to review state-of-the-art development learning-based estimation. We give an overview published papers between 2014 2020 terms training manners types. firstly summarize models Secondly, we categorize various Thirdly, introduce publicly available dataset evaluation metrics. And also analysis properties these compare their performance. Finally, highlight challenges order inform future research directions.