Lightweight Aircraft Door Detection Algorithm Based on Improved YOLOv5

The construction of future smart airports requires automatic docking between boarding bridges and aircraft doors. The important condition for the connection between the aircraft door and the boarding bridge is to accurately measure the distance of the aircraft door, and the core of the distance measurement is to accurately locate the aircraft door. In order to accurately locate and detect aircraft doors, this paper proposes a door recognition and positioning method based on YOLOv5. This article adds CBAM to the ninth layer of the original YOLOv5 network, enhancing the algorithm's feature extraction ability for aircraft doors. In order to improve the detection ability of YOLOv5 for multi-scale targets, this paper adopts BiFPN to improve the detection accuracy of the algorithm. In order to reduce the volume of the YOLOv5s model, this paper adopts a channel pruning method, significantly pruning the honor parameters of the network. The final experimental results show that the improved YOLOv5s model has a volume of 3.66MB, a reduction of 73.3%, an FPS of 370, and a 48% increase in FPS. This obviously reduces obstacles for the subsequent deployment of embedded devices and the achievement of accurate ranging of cabin doors.