Improvement of CNN-Based Model for Object Classification in Aero Photographs

Van Trong Nguyen, Pashchenko Fedor Fedorovich, Bui Truong An, Duc Tiep Le
A review of methods for solving computer vision problems by classes proved the advantages of the neural network method over all those proposed earlier. The principle of operation of neurons in an artificial network makes it possible to identify its activity, which is determined by its parameters for classifying objects in aerial photographs. The article presented a well-founded mathematical model and the main technologies of the convolutional neural network model based on the dropout technique, which is widely used in preventing network retraining. This model helps to increase the invariance to the scale of the input images. At the output of the convolutional layers of the network, several layers of the neural network are additionally installed that perform the work of the classifier.