Vision-based approach for urban vehicle detection & classification

Vehicles detection and classification are the most popular subjects in the computer vision researching field, and also are the most important parts in any traffic monitoring or surveillance system. Although there has been a considerable amount of ideas to accommodate this problem since the 90s, many problems are still unresolved due to the complexity of traffic systems and the variety of vehicles. This paper is a work-in-process that proposes a new approach to detect and classify vehicles based on the traffic system in Vietnam. The main goal of this method is to group vehicles into 2 main classes, which are 2-wheeled and 4-wheeled vehicles, based on low-level traffic parameters in urban areas.