The traditional autonomous vehicle systems require traffic sign detection as a fundamental capability, which allows their real-time detection of road signs, thus improving safety. The objective of this research is to investigates YOLOv9 and YOLOv10 technology as robust traffic sign detection methods for diverse environmental conditions. YOLOv9 enhances learning performance through PGI and GELAN but YOLOv10 surpasses it by utilizing NMS-free training, dual-label assignments along with rank-guided block structure for boosted accuracy, decreased costs and shorter inference times. The YOLOv10 edition surpasses YOLOv9 in traffic sign tests by delivering +1.4% AP gain and operational latency reduction by 65% and parameter-cutting by 36% compared to YOLOv8 which gives it better suitability for real-time applications. An autonomous navigation system with integrated trained models in order to showcase their potential to assist real-time driver operations. YOLOv10 delivers the peak performance for accurate object detection but YOLOv9 provides a suitable alternative for implementing on minimal resource devices. Research on V2X communication systems will advance in order to boost traffic awareness and safety features.
Autonomous Vehicle Navigation and Road Safety by Using the YOLO Algorithm
2025-04-23
729636 byte
Conference paper
Electronic Resource
English
Autonomous Unmanned Vehicle Automatic Visual Tracking Based on SLAM and YOLO Algorithm
Springer Verlag | 2024
|