Application and Optimization of Artificial Intelligence in Smart Car Lights
XU Jian
CHANG Yanan
LIU Yu
Abstract：There are many safety hazards of traditional vehicle lights in night driving,such as fixed light range can not adapt to complex road conditions,easy to cause glare to the opposite vehicle drivers and other problems,seriously affecting driving safety.In this paper,we propose an intelligent headlight control system based on YOLOv8 neural network,which dynamically adjusts the headlight brightness and headlight illumination range by detecting the weather,vehicles and pedestrians on the road in real time to enhance driving safety.In order to improve the detection performance of the network,this paper introduces the attention mechanism and proposes a new loss function for the accuracy of the frame.The experimental results show that,compared with the original model,the improved method reduces the miss mark of target prediction by nearly 5%,and improves the accuracy of prediction frame by more than 23%.The system finally controls the pixel brightness of the headlights through the detection results to realize the accurate obstacle avoidance lighting,which effectively reduces the visual interference while driving.
Keywords：artificial intelligenceintelligent vehicle lightsdigital lightingintelligent manufacturing
Publication Date：2025-02-24
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：7( 26-31,37 )