The traffic in the present is part of a problem to level global, in the countries of Latin America there is a very fast growth in the world. In Perú, the city of Lima maintains one of the more complex traffic congestion in South America and there are few references to growth and a low knowledge of it that do not allow making the right decisions to provide effective solutions, in this sense the project develops an automatic vehicle counting by video in three different points of the city of Lima, to analyze the vehicular flow and thus show a real data of the current situation of the city, this project applies the differentiation of the vehicle with the area without movement by means of adaptive Gaussian mixture, followed by morphological processes for its correct visualization and identification algorithms that will allow labeling vehicles and counting them, an efficiency of 96.9% was obtained in vehicle counting and it was able to give us a vehicle flow analysis of 109 vehicles per minute.)