Enhancing smart city transportation networks: A novel particle swarm optimization and YOLO7x-based ensemble model for accurate vehicle detection

This research incorporates an intelligent transportation system (ITS) of smart cities with a newly developed ensemble model, termed EPYolov7x, through the association of particle swarm optimization (PSO) and YOLOv7x for high-accuracy traffic congestion prediction. In the proposed model, PSO is used for hyperparameter optimization for the YOLO model. Experimental results suggest a mean average precision (mAP) of 98.6% with an inference speed of 106 FPS on the UA-DETRAC dataset. It also provides good precision and real-time performance; both are essential for the practical deployment of ITS. The study findings confirm that PSO-based hyperparameter optimization improves the detection accuracy for complex traffic scenarios. The obtained results indicate superior performance compared to existing approaches. The proposed model can facilitate sustainable urban mobility toward a safer future with optimal computation.