Vehicle Detection Counting and Speed Estimation
Abstract: This project introduces a sophisticated vehicular monitoring system that leverages deep learning for real-time vehicle detection, classification, and speed estimation. The implementation is based on the YOLOv8 Nano model for high-performance object recognition, focusing on dynamic vehicle classes such as automobiles, trucks, and buses. A custom tracking algorithm is employed to maintain continuous vehicle trajectory tracking across video frames. Speed estimation is computed through pixel-based distance metrics and temporal analysis, applying distance-time calculations for precise measurements. DenseNet layers, coupled with dropout mechanisms, are utilized to enhance detection accuracy and generalization. The system effectively identifies vehicles violating speed regulations, providing a robust tool for intelligent traffic management, real-time surveillance, and road safety enforcement. Experimental results validate the system's reliability in automating vehicle tracking, velocity estimation, and overspeed detection within traffic monitoring infrastructures.
Keywords: Vehicle detection, Speed estimation, Traffic monitoring, Deep learning surveillance, Overspeed detection, Vehicle classification, YOLOv8 Nano model, Object recognition, Vehicle tracking
How to Cite?: Sohail, Dr. Satish Kumar Satti, Gayathri, "Vehicle Detection Counting and Speed Estimation", Volume 15 Issue 5, May 2026, International Journal of Science and Research (IJSR), Pages: 1476-1480, https://www.ijsr.net/getabstract.php?paperid=SR26521085403, DOI: https://dx.doi.org/10.21275/SR26521085403