Traffic Target Integrated Detection and Spatial Visualization System Based on Dual YOLOv11 Parallel Architecture

Traffic Target Integrated Detection and Spatial Visualization System Based on Dual YOLOv11

Abstract—To address the challenges of insufficient collaborative detection accuracy for traffic signs and vehicles, weak spatial visualization, and the imbalance between system integration and lightweight deployment in intelligent transportation scenarios, this paper proposes a traffic-target integrated detection and spatial visualization system based on a dual YOLOv11 parallel architecture. Following the technical route of independent training, parallel inference, and result fusion for dedicated traffic-sign and general-purpose vehicle models, the system incorporates frame-rate-based adaptive sampling, three-level spatial coordinate mapping, and digital-twin canvas rendering. It realizes a full-process closed loop encompassing multi-source input, dual-model parallel detection, lightweight video processing, spatial visualization, interactive control, data statistics and export, as well as user authentication and permission management. Experimental results demonstrate that the system achieves a traffic-sign detection mAP@0.5 of 92%, a comprehensive detection rate of 90%, an overall vehicle detection mAP of 92.7%, and a single-frame inference time of approximately 0.28 s, meeting real-time requirements. With the adaptive sampling strategy of one frame per second, CPU usage decreases by about 65% while processing efficiency increases threefold. Built with PyQt5 and SQLite, the system features convenient deployment, cross-platform compatibility, and strong scalability for applications in intelligent traffic management, driver assistance, traffic digital twins, flow monitoring, and teaching experiments.

Keywords—Yolov11; Dual-Model Parallelism; Traffic Sign Detection; Vehicle Detection; Spatial Visualization; Digital Twin; Intelligent Transportation; Pyqt5