Detection and Recognition of Road Information and Lanes Based on Deep Learning

Abstract

During the driving process, it is essential for the acquisition of road information around the vehicle, and it is also an indispensable part of the autonomous driving assistance system (ADAS). The overall ADAS system can be divided into perceptual layers, decision-making layers, and execution layers, while the core is to carry out environmental perception. This article proposes a road traffic symbol based on deep learning and a lane detection identification framework. This framework uses a monocular camera to collect the driving environment information in front of the vehicle, combining the improved YOLOV4 algorithm with the LaneNet lane detection algorithm, Testing and identifying traffic signs, transportation lights, vehicles, pedestrians, riders and lanes, and realized the visual perception part of unmanned cars. The experimental results show that the framework proposed in this article can accurately detect roads and lane information during driving, and has certain advantages in detection accuracy.