Airport departure layer violation vehicle automatic identification method based on deep learning

The method comprises the following steps: based on a monitoring video of an actual fixed camera position on an airport departure lane side, using a YOLOv4 algorithm to identify vehicle and pedestrian targets, and using a DeepSORT algorithm to track the targets; and the running state of the vehicle and the getting-on and getting-off behaviors of passengers are judged according to the target category returned by the YOLOv4 and the target position information recorded by the DeepSORT, the change condition of the number of passengers in the vehicle is counted when the vehicle stops, whether the vehicle has the passenger receiving behavior or not is judged according to the statistical result, and recognition of the violation vehicle on the departure floor of the airport is completed. The method has the beneficial effects that the accuracy can reach 83.3%, the detection speed is greatly improved compared with a traditional manual identification mode, the workload of monitoring law enforcement personnel can be great