Special Lane Detection Based on YOLOv3-SPP

As one of the key technologies of driverless driving, lane detection has certain challenges. With the development of deep learning, many lane detection algorithms based on deep learning have been proposed. Under construction road conditions, continuous obstacles will be placed to guide the vehicle, at this time, the continuously placed obstacles constitute a special lane. In order to identify this special lane line, this paper proposes a lane detection method based on Yolo. First, the algorithm uses an object detection neural network to identify the location of obstacles in the picture, and then uses the least squares method to fit lanes. Compared with the previous work, this paper proposes the concept of special lane lines and a corresponding detection method. Finally, the validity of the detection algorithm proposed in this paper is verified by experiments.