Traffic Measurement and Authomatic Vehicle Classification from Videos
Computer vision, deep learning, traffic engineering, public safety
Studies show that the trend of personal vehicles is the increase in their quantities due to populational growth of big urban centers. Thus, the problem traffic jams is one of the greatest challenges facing these centers. Moreover, with a big number of vehicles circulating in the cities, we also have an increase in the number of vehicle thefts. Looking at this problem, we propose a system for helping public organizations to perform strategic planning using low cost components, such as CFTV cameras, to, through video processing, extract vital data for this planning. The proposed system has three modules that are responsible for segmenting, tracking and recognizing/classifying vehickes in videos using computer vision techniques and deep learning. We intend to use traffic monitoring cameras to extract, in real time, informations for the public officces responsible for urban planning ans public safety.