Application of deep learning and unmanned aerial vehicle technology in traffic flow monitoring

Intelligent video surveillance technology has been increasingly used in the field of transportation. Real-timely capturing traffic video data through the UAV is a new way to get road condition. In this paper, we set the statistics of road traffic flow as the starting point. After analyzing the characteristics of videos shot by the UAV, we choose to use the deep learning framework based on Faster-RCNN to train a vehicle detection model to detect vehicle targets in videos. The motion track of vehicles in the shooting scene were drawn according to the result of object detection. In the end, analyzing the track and calculating the traffic flow. From the experimental results, it can be seen that deep learning method can achieve a high detection accuracy and based on this, we can calculate the traffic flow well.