Mask R-CNN Algoritması ile Hangar Tespiti Hangar Detection with Mask R-CNN Algorithm

Abstract

In this study, the detection of hangars in high resolution airport (civil and military) satellite images was performed using Mask R-CNN algorithm. Although the detection of buildings in the satellite images is a common practice, being some of the hangars camouflaged in different sizes cause difficulty for the detection algorithms. In Mask R-CNN, an instance object segmentation algorithm, objects in the images are detected, bounding box of each object as well as their pixel information with in the box are marked separately. In this study, high resolution hangar data set with 300 samples, collected from various air bases, was prepared, and an 85% average precision is achieved using Mask R-CNN.