In computer vision, optical flow plays a crucial role in object tracking and velocity estimation. However, traditional optical flow methods often suffer from noise, occlusions, and irrelevant motion, limiting their effectiveness in dynamic environments. This paper presents an enhanced optical flow tracking framework that integrates motion saliency to improve tracking accuracy and robustness. A novel saliency-based weighting mechanism is introduced, which dynamically adapts key-point recalibration and prioritizes motion-relevant regions. In busy environments, this improves feature tracking and lessens distractions. The framework works well in a variety of traffic situations, such as multilane highways and crowded areas, according to experimental data. In addition to producing motion saliency maps and accurately estimating speeds within ±10% of ground truth values, the system is able to track vehicle motion. Near real-time capabilities are demonstrated by the computational pipeline, which on normal hardware produces a processing rate of about 15 frames per second (FPS). The aforementioned findings demonstrate the appropriateness of the suggested approach for uses including traffic monitoring, autonomous navigation, and intelligent transportation systems.