Application of Vehicle Detection Based On Deep Learning in Headlight Control

Abstract

When driving at night, vehicle lights are the greatest guarantee for driving safety. Drivers often turn on the high beams to make the oncoming vehicle unclear, turn on the high beams of the oncoming vehicle to reduce their visual range or turn on the high beam when driving behind. The lights prevent themselves from being able to check the rear view through the rear mirror and cause traffic accidents. In addition, drivers often fail to keep a safe distance from the traffic, which causes serial accidents. In response to this problem, this paper proposes a deep learning-based image recognition in the headlight control system. Vehicles using this system can detect the vehicle ahead in real time when the driver is driving the vehicle, and calculate the safety distance of the vehicle ahead, and judge at night Whether there are vehicles in the front and oncoming lanes to determine whether to turn on the high beam, so as to reduce light damage and safe distance traffic accidents.