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.
MiNet: A Convolutional Neural Network for Identifying and Categorising Minerals
Authors:
Emmanuel Asiedu Brempong,
Millicent Agangiba,
Daniel Aikins
Abstract:
Identification of minerals in the field is a task that is wrought with many challenges. Traditional approaches are prone to errors where there is no enough experience and expertise. Several existing techniques mainly make use of features of the minerals under a microscope and tend to favour a manual feature extraction pipeline. Deep learning methods can help overcome some of these hurdles and prov…
▽ More
Identification of minerals in the field is a task that is wrought with many challenges. Traditional approaches are prone to errors where there is no enough experience and expertise. Several existing techniques mainly make use of features of the minerals under a microscope and tend to favour a manual feature extraction pipeline. Deep learning methods can help overcome some of these hurdles and provide simple and effective ways to identify minerals. In this paper, we present an algorithm for identifying minerals from hand specimen images. Using a Convolutional Neural Network (CNN), we develop a single-label image classification model to identify and categorise seven classes of minerals. Experiments conducted using real-world datasets show that the model achieves an accuracy of 90.75%.
△ Less
Submitted 22 November, 2021;
originally announced November 2021.
