Research on Real-Time Detection and Anti-Interference of Road Traffic Signs Based on YOLOv8 Algorithm in Automatic Driving Environment

Abstract

In the stage of rapid development of automatic driving technology, the core link to ensure driving safety focuses on the accurate real-time detection of traffic signs, however, the complex reality of environmental interference factors such as sudden changes in lighting, dynamic occlusion, bad weather and hardware noise greatly limits the performance of traditional detection methods. In this paper, we focus on the YOLOv8 algorithm and consider its optimization potential in the field of real-time traffic sign detection and anti-interference capability. The study shows that the existing methods have improved the detection accuracy by improving the feature fusion mechanism and introducing the attention module, but they still have the defects of lack of model generalization ability, high demand of computational resources and low stability of small target detection, etc. Therefore, we propose the lightweight design, dynamic environment adaptation optimization and anti-jamming module fusion strategy for YOLOv8. YOLOv8 greatly improves detection efficiency and accuracy with its anchorless detection mechanism, upgraded C2f backbone network, and task decoupled detection head design, while its integration of multi-scale feature extraction, data enhancement, and regularization techniques addresses interference problems in complex environments. YOLOv8 shows excellent performance in conventional scenarios, but there are still ups and downs in extreme weather and severe occlusion situations. This paper provides theoretical support and innovative tendency for the engineering application of traffic sign detection technology in the context of autonomous driving.