Accurately monitoring pedestrian and vehicular flow is crucial for traffic management, urban planning, and public safety. Conventional approaches fail to provide greater performance in both accuracy and efficiency. This study used a deep learning object identification model like YOLOv9 to recognize and track pedestrians and cars in real-time video streams. Furthermore, this methodology goes beyond simple detection by integrating speed estimation capabilities. Examining consecutive frames and applying motion estimation techniques helps to estimate the speed of found vehicles precisely, so augmenting our knowledge of urban dynamics. Apart from counting the cars and pedestrians crossing the camera, it can recognize vehicles and objects. Results of tests on benchmark datasets show the value of the suggested method, which surpasses present methods in terms of accuracy, precision, recall, and mean average precision.