Automatic Detection of Ambulance Vehicles in Day and Night Conditions in Surveillance Videos

Analysis of surveillance videos is one of the most important task in the computer vision area. Street traffic monitoring is useful for developing the traffic light systems at road intersections, detection of red light violations, traffic intensity study, optimization of traffic organization, and also for ensuring the secure interaction of autonomous vehicles with emergency vehicles. The detection of emergency vehicles mainly such as ambulances, fire trucks, and police cars with flashing lights and sirens, but also some others like tow trucks or snow plows is frequently based on the analysis of vehicle shape or color as well as sound of sirens. The observations and the initial tests have shown that the detection of ambulances in surveillance videos recorded during the day is successful if it is mainly based on the acoustic analysis. Whereas, for surveillance videos recorded at night or late in the evening a more efficient approach is the analysis of rotating and flashing lights installed on ambulances.