Wojciechowski, WitoldNowakowska, Leslawa M.Zajac, Zofia G.Kaczmarek, Walenty J.Boz, Ilayda2025-03-172025-03-172024979-833153149-2https://doi.org/10.1109/IDAP64064.2024.10710941https://hdl.handle.net/20.500.13099/14008th International Artificial Intelligence and Data Processing Symposium, IDAP 2024 -- 21 September 2024 through 22 September 2024 -- Malatya -- 203423Advancements in technology have spurred the development of autonomous driving systems, necessitating new innovations to enhance road safety and address transportation challenges. This study evaluates and compares the performance of YOLOv7 and YOLOv8 models from the YOLO family in detecting traffic signs, which are crucial for road safety. Using a comprehensive dataset sourced from the Google Earth platform, the research examines the limitations and successes of both models. YOLOv8 demonstrates notable performance in traffic sign detection under challenging conditions, with precision, recall, and mAP values of 0.866,0.815, and 0.807, respectively. While YOLOv7 shows competitive results, it falls short of YOLOv8 in difficult scenarios. The study provides a detailed analysis of both algorithms, exploring their strengths, weaknesses, and areas for potential improvement in traffic sign detection. © 2024 IEEE.eninfo:eu-repo/semantics/closedAccessdeep learningobject detectionTraffic sign detectionYOLOv7YOLOv8Conference Object10.1109/IDAP64064.2024.107109412-s2.0-85207936850N/A