International Journal of Progressive Research in Engineering Management and Science
(Peer-Reviewed, Open Access, Fully Referred International Journal)
email@example.com or Whatsapp at (+91-9098855509)
REAL TIME TRAFFIC SIGNAL SWITCHING USING MACHINE LEARNING (KEY IJP************839)
ABSTRACTTraffic clog is the significant issues in India and is particularly prevalent in the metropolitan urban areas of the nation. This clog increases day by day on the road as the vehicle density increases, so managing such large traffic nowadays through traditional approach is not so successful. The traffic is larger on one side than the other side in such case the traditional approach fails. So, in this project we perform real-time signal monitoring and handling approach (i.e.,) moving from static switching to signal switching. We use an algorithm YOLOv3 (You Only Look Once, Version 3) which is a real-time object detection algorithm that identifies specific objects in videos, live feeds, or images, to find the vehicle count in two side of the roads and perform signal switching based on the count. This count is fed as input to raspberry pi 3 to perform signal switching and these uses led lights to toggle the signal. The switching time of signal will be decided based on detection of real time image with accurate results even in dense traffic. Overall, in our project we compare the object counts from the different camera and perform signal switching dynamically. to perform signal switching and these uses led lights to toggle the signal. The switching time of signal will be decided based on detection of real time image with accurate results even in dense traffic. Overall, in our project we compare the object counts from the different camera and perform signal switching dynamicallyI.INTRODUCTIONPeople in todays era usually have tendency of using their own vehicles for commutation rather than using public or pooled means of transport and this results in large number of private vehicles on road. This endless increasing number of vehicles on road gives rise to many problems amongst them traffic congestion tops in every aspect. In such scenario one cannot restrict individual to limit the usage of the private vehicles but what we can do is at least manage traffic flow in a way that it doesnt alleviate congestion issues.There are projects emerging in order to convert the current transport system of cities to Smart system and there are various initiatives under this, one of the initiatives is Intelligent Transport System. Many initiatives were taken to design a system that can perform real- time monitoring of signals during the traffic i.e. the signal switching time of traffic will not be predefined one, instead the switching time will depend on the number of vehicles on either side of the road. This process of getting the number of vehicles on the road can be achieved using various detection techniques.Our aim is to design and develop a miniature to depict the current road situation along with monitoring and handling the traffic issues. Hence to proceed with this project we are using a pre-trained model YOLO to perform the task of object detection. The pretrained model YOLO uses OpenCV for object detection of object along with multiple background and foreground subtraction and removal of noise from the input image. The CCTV cameras that are used for surveillance purpose can be made use to capturing the footage of the road, this image will be passed to the pretrained model as input image. To do so each side of the road will be divided into particular frames of same height and width for capturing the image. The count or number obtained from the image is fed to the Raspberry board. As per the count obtained, switching time will be assigned for either side of road. The program will initially check if the number of vehicle in all frame is approximately same then the switching will remain at its predefined regular interval for all sides of signal, the real- time switching for the signal will be performed if the number of vehicles in all fames varies as threshold difference which be provided.