The well-grounded discernment of speed of moving vehicles is considered solution to traffic law implementation in most countries, and is seen by many as a key tool to decrease the number of road accidents and fatalities. Many automatic systems and many methods are engaged in different areas, they tend to be exorbitant and/or labor intensive, frequently employing old-fashioned technology due to the long progress time. Here we report a speed detection system that depends on simple everyday equipment - a laptop and a web camera. The proposed system relies on tracking the cars, which gives the track of the car using respective algorithm with the help of recorded video stream. In the thesis part, the proposed system deals with the technology such as Adaptive background subtraction based on the method Gaussian Mixture Model, and DBSCAN, the clustering algorithm for forming the clusters and the Kalman fillter for tracking the selected automobile. Consequently, our system evaluates the actual speed of moving vehicles precisely. The output of the system shows promising results on videos obtained in a lot of scenes and with different templates.