[3F1-ES-2] Machine learning: Social application (3)

[3F1-ES-2-02] Prediction of Lane Number by Combining Results From Lane Detection

〇Panumate Chetprayoon¹, Fumihiko Takahashi¹ (1. JapanTaxi Co.,Ltd.)

キーワード：Machine Learning, Deep Learning, Computer Vision, Lane Detection, Road Scene Understanding

The lane number that the vehicle is travelling in is one of an important factors in intelligent vehicle fields. Nowadays, there are so many novel lane detection algorithms proposed. If we can perfectly detect the lanes, we can directly calculate the lane number from the lane detection results. However, in some conditions such as adverse weather conditions or vehicle occlusions, these lane detection algorithms cannot perform well. Therefore, this paper proposed a new approach for predicting the lane number that the vehicle is currently travelling in from drive recorder video data. We came up with the idea that we could combine the drive recorder image with the lane detection result, which is sometimes imperfect, to predict the lane number. Experiments on our own dataset confirmed that our approach consistently delivers outstanding results.