| Abstract |
"In the pursuit of advancing real-time speed estimation and vehicle detection, this paper introduces a novel methodology that combines Kalman Filter and Single Shot Multi-box Detector (SSD) techniques. Tailored for tasks such as traffic monitoring, the Kalman Filter dynamically estimates and forecasts vehicle speeds, taking into account their dynamic motion patterns. This adaptive filtering mechanism iteratively updates speed estimations, ensuring robustness in tracking vehicles with diverse velocities and trajectories. Concurrently, SSD efficiently identifies vehicles in video frames, offering multiple bounding boxes and class probabilities. This collaborative strategy not only improves the accuracy and effectiveness of tracking vehicles with varying speeds but also holds potential for substantial contributions to intelligent transportation systems and smart city initiatives. The model achieves a balanced F1 score of 95.84%, demonstrating its high precision (97.12%) and recall (94.61%). Moreover, it accurately detects and predicts vehicle speeds with an overall accuracy of 87.26%, highlighting its practical utility and potential for further advancements in intelligent transportation technologies." © 2024 IEEE. |
