Neural network learning to identify airport runway taxiway numbers

In order to reduce the number of 'Runway Incursion' crashes, the process and method of the study is to simulate human visual neuron recognition pictures and train a machine learning model. The results are modeled as follows:1. The neural network outputs the feature vectors model after the Convolution operation. 2. Filter is used to detect the pixels of the background image of the airport environment. The feature map is output with Activated feature map. The same weight of the same feature object is added with Max pooling layer after convolution layer to find the feature map more obvious. We can use Deep Learning to simulate the Artificial Intelligence model environment instead of manual recognition, simulate Drone to recognize the runway number automatically and drive into the tower path planning runway area correctly. Specific images such as runway numbers, area locations, etc., provide an unexpectedly increased opportunity to resolve the 'Runway Incursion' case.