<p>Globally, traffic congestion impedes economic development. Dhaka, the fifth most crowded city in the world, faces significant traffic problems as a result of a lack of an intelligent monitoring and forecasting system. The lack of an efficient solution leads to unmanageable congestion, longer travel times, and economic losses. To address this issue, we propose a real-time traffic monitoring system capable of vehicle detection, counting all vehicle types, speed estimation, and traffic forecasting. A custom vehicle image dataset for Dhaka city was created by extracting images from videos of different road segments to build a robust model. This dataset is used with YOLOv8 (You Only Look Once) for vehicle detection, which quickly identifies objects in camera footage. Training YOLOv8 on five vehicle types in Dhaka city (truck, bus, car, motorcycle, and three-wheeler) achieves around 94% accuracy. The system can count vehicles and measure their speed in real time using very low-resolution cameras. Vehicle counts and average speeds are then collected to form a numerical dataset. This dataset trains machine learning models to predict traffic flow and aid traffic management decisions. Vehicles are tracked frame-by-frame with distance and time data recorded to calculate average speeds. Using LSTM, we achieve 83% accuracy for predicting average speed and 97% accuracy for predicting vehicle count. These insights enable authorities to optimize signal timings, reduce congestion, and improve overall road efficiency, ultimately leading to smoother transportation, reduced travel time and enhanced economic productivity.</p>