Research And Application Of Foreign Object Debris Detection In Airport Runway Based On Convolution Neural Network

With the development of economy,the proportion of air transportation and air travel is increasing.When the aircraft takes off or lands,any foreign object on the runway of the airport will bring great threat.At present,many airports in China detect foreign matters on the runway by means of manual inspection.At present,the technologies used in the airport runway foreign body detection system include radar based detection technology and optical image based detection technology.Over the years.With the rapid development of deep learning technology,the target detection algorithm based on convolutional neural network has made great progress.Some objects in the airport runway are relatively small,which may not be detected by the existing algorithm directly,resulting in low detection accuracy.In order to solve this problem,this paper studies and applies the algorithm of foreign body detection in airport runway based on convolution neural network,the specific work is as follows:In order to solve the problem of small object difficult to detect in airport runway,based on the research of high-resolution network,this paper constructs a highresolution network to detect airport runway foreign objects.Here,by parallel connection and fusion of high-resolution convolution and low-resolution convolution,high-resolution representation can be maintained in the process of convolution.Through the multi-level average pool sampling,the multi-level characteristic map is constructed.The high-resolution network is used as the backbone network in other algorithms,and experiments are carried out on the FOD3(Foreign Object Debris Detection Dataset,FODDD)dataset.The results show that the use of high-resolution network can effectively improve the detection accuracy of the algorithm for small object in FOD.On the basis of using high-resolution network as the backbone network,this paper studies and analyzes the setting of the threshold value of IoU(Intersection over Union)in the target detection algorithm,adopts the idea of cascading,and improves the detection accuracy of the network by cascading different IoU thresholds,so as to improve the detection accuracy of the algorithm for airport runway foreign matters.Experiments on FOD3 dataset show that adding the structure of IoU threshold cascade can improve the detection effect of the algorithm for the object in FOD.The algorithm of foreign object debris detection in airport runway based on global context information is designed and implemented.In the process of capturing image global context information by the original Non-local Block,the global context information obtained by each location is almost the same,so it is simplified to replace other locations with one location.Then,the SE(Squeeze-and-Excitation)block is integrated to model the relationship between feature channels.Finally,the global context block is formed.The global context block is added to the backbone network of the existing algorithm and the experiment is carried out on the FOD3 dataset,which proves that the addition of the global context block can improve the detection performance of the algorithm.