PROJECT TITLE :

Vehicles Detection for Smart Roads Applications on Board of Smart Cameras A Comparative Analysis

ABSTRACT:

The use of video analytics in smart roads environments can be lucratively adopted for the purpose of automatically detecting abnormal circumstances. In this context, the first and most important stage is vehicle detection, and it is essential that this stage be accurate because any mistake in vehicle detection will affect the performance of any step that comes after it. Additionally, in smart road environments, it is frequently preferred to perform the video analysis directly on board of smart surveillance cameras. This is done in order to decrease bandwidth usage and eliminate the cost of setting up and maintaining powerful processing servers. On the other hand, processing on board of smart cameras implies that the detection algorithm must be quick and slim because the resources available on this kind of embedded device are limited. In this day and age of deep learning, it would appear that the question of which method is the most effective one for detecting vehicles may have an easy answer, given that this category of methods includes some that are exceptionally precise. In any case, according to the considerations presented above, the method that is best suited for this application is not necessarily the most accurate one, but it is without a doubt the most accurate one that can run on the hardware that is currently available at a particular frame rate and resolution. In this paper, we perform an analysis of the methods available in the literature for vehicle detection, by comparing them in terms of accuracy and computational burden, with the aim of answering the following question: what is the best method for vehicles detection when working with smart cameras? Starting from the considerations presented above, we perform this analysis in order to answer the question.