High-Resolution Building and Road Detection from Sentinel-2
Authors:
Wojciech Sirko,
Emmanuel Asiedu Brempong,
Juliana T. C. Marcos,
Abigail Annkah,
Abel Korme,
Mohammed Alewi Hassen,
Krishna Sapkota,
Tomer Shekel,
Abdoulaye Diack,
Sella Nevo,
Jason Hickey,
John Quinn
Abstract:
Map** buildings and roads automatically with remote sensing typically requires high-resolution imagery, which is expensive to obtain and often sparsely available. In this work we demonstrate how multiple 10 m resolution Sentinel-2 images can be used to generate 50 cm resolution building and road segmentation masks. This is done by training a `student' model with access to Sentinel-2 images to re…
▽ More
Map** buildings and roads automatically with remote sensing typically requires high-resolution imagery, which is expensive to obtain and often sparsely available. In this work we demonstrate how multiple 10 m resolution Sentinel-2 images can be used to generate 50 cm resolution building and road segmentation masks. This is done by training a `student' model with access to Sentinel-2 images to reproduce the predictions of a `teacher' model which has access to corresponding high-resolution imagery. While the predictions do not have all the fine detail of the teacher model, we find that we are able to retain much of the performance: for building segmentation we achieve 78.3% mIoU, compared to the high-resolution teacher model accuracy of 85.3% mIoU. We also describe a related method for counting individual buildings in a Sentinel-2 patch which achieves R^2 = 0.91 against true counts. This work opens up new possibilities for using freely available Sentinel-2 imagery for a range of tasks that previously could only be done with high-resolution satellite imagery.
△ Less
Submitted 20 June, 2024; v1 submitted 17 October, 2023;
originally announced October 2023.
