Paper Title
Highway Traffic Surveillance By Unsupervised Learning

Abstract
This paper presents an approach to a fine method to detect vehicles in video sequences. The detection process is divided into two stages: ROI (region of interest) generation stage and ROI classification stage. In the first stage Haar-like features are exploited to rapidly search the whole image and find interesting regions which may contain the vehicle. This method incorporates well-studied computer vision and machine learning techniques to form an unsupervised system, where particular vehicles (cars, SUVs and heavy vehicles) are automatically “learned” from video sequences. An adaptive background mixture model is used to identify the positive vehicles; then a classifier with trained examples is used for detection wherein both background subtraction and the classifier are used to achieve very accurate results while not compromising efficiency. The proposed method exploits robust vehicle detection to allow for particular vehicle tracking in high density traffic. This method is tested for Mumbai’s Roads (including National Highways NH-8 and NH-17, regular roads as SV road and Link road) in various situations as for day time, night time, sunset and under high traffic. The proposed surveillance is also tested in low, medium and high resolution with high level of accuracy and precision. The proposed model is giving an acceptable and high range of detection accuracies from 87% to 94%