2026-04-19T22:05:58Z https://ipsj.ixsq.nii.ac.jp/oai

oai:ipsj.ixsq.nii.ac.jp:00241873 2025-01-19T07:30:50Z 6164:6165:9654:11891

Identifying Traffic Direction via YOLOv8 and Cross Product Method Identifying Traffic Direction via YOLOv8 and Cross Product Method Metawin, Sumethiwit Chinpakorn, Waiyavudhi Daranee, Hormdee Metawin, Sumethiwit Chinpakorn, Waiyavudhi Daranee, Hormdee This paper presents a novel approach to identifying traffic direction using a combination of the YOLOv8 object detection algorithm and a cross product-based method for vector analysis. YOLOv8, a state-of-the-art real-time object detection model, is employed to accurately detect and track vehicles in video streams. The detected vehicles are then analyzed using a Cross Product method, which calculates the relative movement vectors of the vehicles across consecutive frames. By determining the orientation of these vectors, the traffic direction―whether vehicles are moving left, right or forward―is inferred. This approach offers a robust solution for real-time traffic monitoring and analysis, leveraging the efficiency of YOLOv8 for object detection and the mathematical precision of the cross product for movement analysis. The proposed method is tested on various traffic scenarios, demonstrating its effectiveness in identifying traffic direction under diverse conditions. This paper presents a novel approach to identifying traffic direction using a combination of the YOLOv8 object detection algorithm and a cross product-based method for vector analysis. YOLOv8, a state-of-the-art real-time object detection model, is employed to accurately detect and track vehicles in video streams. The detected vehicles are then analyzed using a Cross Product method, which calculates the relative movement vectors of the vehicles across consecutive frames. By determining the orientation of these vectors, the traffic direction―whether vehicles are moving left, right or forward―is inferred. This approach offers a robust solution for real-time traffic monitoring and analysis, leveraging the efficiency of YOLOv8 for object detection and the mathematical precision of the cross product for movement analysis. The proposed method is tested on various traffic scenarios, demonstrating its effectiveness in identifying traffic direction under diverse conditions. conference paper 情報処理学会 2024-12-27 application/pdf Proceedings of Asia Pacific Conference on Robot IoT System Development and Platform 2024 57 58 https://ipsj.ixsq.nii.ac.jp/record/241873/files/IPSJ-APRIS2024013.pdf eng