Examination of How to Estimate the Viewing Angle at Intersections
Shinichiro Goto, Yuki Murata, Masayasu Atsumi · 2018
We are considering a system capable of providing a route of minimum driving risk to reduce accidents of elderly drivers. In this system, mapping the driving risk is an important issue so that it is necessary to estimate the risk level of intersection without traffic signal where accidents occur frequently. To estimate the risk level of intersections, the viewing environment, especially the viewing angle, has a big influence. We plan the automatic generating method to predict the viewing angle from the Google Street View using latest computer vision approach. We try to predict the viewing angle to recognize the shield objects by using Semantic Segmentation.