Traffic Flow Prediction using Yolo Algorithm

Traffic flow prediction is very important in urban regions were traffic congestion is a major problem. The traffic congestion create many problems such as wastage of time, carbon emission leads to pollution, traffic blocks which can lead to many other problems. Hence by using traffic flow prediction methods we can reduce these problems. Using this methods we can predict the situation of traffic at the lane and we can plan our journey accordingly. In our proposed system we develop a mobile application which is platform independent which is used for the prediction of traffic in a region. In our system it is real time monitoring system where our input is the video taken from the road. Those video will be passed to YOLO (You Only Look Once) algorithm for object detection. The algorithm use a neural network to the full image, and then divides the image into grids and predicts bounding boxes and probabilities. The Yolo algorithm use Common Objects in Context (COCO) dataset which has 80 objects including car, bus, person and so on. We train the model such that vehicles will be detected from the video. We take the count of the vehicle and predict the traffic in the region whether low, medium or heavy. Our proposed method is more accurate than other methods since the algorithm can detect static vehicles and also ignore the shadows and reflections.