Pedestrian Detection, Tracking, Counting, Waiting Time Calculation and Trajectory Detection for Pedestrian Crossings Traffic light systems

Road safety and traffic congestion are major concerns when it comes to Sri Lankan context. Pedestrian crossing plays a major role in this. Currently, in Sri Lanka different methods for control traffic lights in pedestrian crossings are available. Such as time base traffic lights and button push traffic lights. Even though these lights help up to some extent to increase safety and traffic flow it is not efficient. Therefore, in this paper it will be discussed how the image processing technologies can be used to pedestrian detection, tracking, counting, waiting time detection and trajectory detection can be done to improve traditional traffic light system. Haar-cascade classifier which used to use detect pedestrians has 50.47% of precision and 85.5% recall. Pedestrian counting module has 72% accuracy and the waiting time calculation model has an average error rate of 0.714. As well as the trajectory detection module has a 78.234% correct prediction rate.