Computer vision: depth estimation from inverse perspective mapping using deep neural networks

In recent years, for the development of autonomous driving technology, it has become important to detect the distance between a vehicle and another object. There are several ways to detect distance, but we try to approach it by replacing visual information. We develop a method to detect the distance between objects by creating a bird-eye’s view image based on the image taken from the front of the vehicle. Through simulation data and deep learning, we try to solve the problem that was impossible due to the lack of spatial information. Our method has different visual advantages such as spatial perception compared to conventional methods like depth map prediction. When our research is activated, it can be used not only in autonomous driving, but also in fields that can acquire visual information by transforming other images.

Sung Jun Ahn, Mijin Park, Sungkyu Bang, Eunseo Cho, Sung Gwe Ahn, Sang Hyun Suh, Jong-Min Lee*, Apparent diffusion coefficient histogram in breast cancer brain metastases may predict their biological subtype and progression, Scientific Reports, Vol.8(1), 9947, July 2018