The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Download
Publications Copernicus
Download
Citation
Articles | Volume XLIV-4/W3-2020
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLIV-4/W3-2020, 215–218, 2020
https://doi.org/10.5194/isprs-archives-XLIV-4-W3-2020-215-2020
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLIV-4/W3-2020, 215–218, 2020
https://doi.org/10.5194/isprs-archives-XLIV-4-W3-2020-215-2020

  23 Nov 2020

23 Nov 2020

BUILDING DETECTION FROM SAR IMAGES USING UNET DEEP LEARNING METHOD

R. A. Emek1 and N. Demir2 R. A. Emek and N. Demir
  • 1Akdeniz University, Institute of Science and Technology, Remote Sensing & GIS Graduate Program, 07058 Antalya, Turkey
  • 2Akdeniz University, Faculty of Science, Dept. of Space Science and Technologies, 07058 Antalya, Turkey

Keywords: SAR, U-Net, Deep Learning, Object Detection, Convolution Neural Network, Radar, Building

Abstract. SAR images are different from the optical images in terms of image properties with the values of scattering instead of reflectance. This makes SAR images difficult to apply the traditional object detection methodologies. In recent years, deep learning models are frequently used in segmentation and object detection purposes. In this study, we have investigated the potential of U-Net models for building detection from SAR and optical image fusion. The datasets used are Sentinel 1 SAR and Sentinel-2 multispectral images, provided from ‘SpaceNet 6 Multi Sensor All-Weather Mapping’ challenge. These images cover an area of 120 km2 in Rotterdam, the Netherlands. As training datasets 20 pieces of 900 by 900 pixel sized HV polarized and optical image patches have been used together. The calculated loss value is 0.4 and the accuracy is 81%.