Investigation of the applicability of the convolutional neural network U-Net to a problem of segmentation of aircraft images

The paper investigates the applicability of convolutional neural network "U-Net" to a problem segmentation aircraft images. image method is based on "Carvana" implementation with architecture. For orientation recognition, built in Keras open library pretrained VGG16 used. approach considered allows be conducted. results experiments have shown possibility fairly accurate selection object interest. resulting binary masks make it possible visually classify image.