Estimation of Lane-Level Traffic Flow Using a Deep Learning Technique

This paper proposes a neural network that fuses the data received from camera system on gantry to detect moving objects and calculate relative position velocity of vehicles traveling freeway. information is used estimate traffic flow. To flows at both microscopic macroscopic levels, this YOLO v4 DeepSORT for vehicle detection tracking. The number passing freeway was then calculated by drawing virtual lines hot zones. each also recorded. can be passed control center in order monitor freeways analyze conditions.