|
snippet:
|
The 2D building footprints are automatically extracted from the 2023 LiDAR-derived nDSM and high-resolution aerial imagery dataset as a complementary dataset to the SL_CGIS_2D_BLD_FTPR. Manual quality assurance is performed on the dataset. |
|
summary:
|
The 2D building footprints are automatically extracted from the 2023 LiDAR-derived nDSM and high-resolution aerial imagery dataset as a complementary dataset to the SL_CGIS_2D_BLD_FTPR. Manual quality assurance is performed on the dataset. |
|
accessInformation:
|
Geomatics Branch, Applications | Information Systems and Technology Department, 1st floor Keller House, 121 Loop Street, Cape Town, 8001
Data distributed by City Maps public counter, email: City.Maps@capetown.gov.za. |
|
thumbnail:
|
|
|
maxScale:
|
5000 |
|
typeKeywords:
|
[] |
|
description:
|
<DIV STYLE="text-align:Left;"><DIV><DIV><P><SPAN>The 2D building footprint dataset is a 2D representation of building roof extents and should not be confused with the building industry floor footprint. This layer should be used to complement the higher accuracy and detail “2D Building Footprint” dataset (SL_CGIS_2D_BLD_FTPR), which is directly derived from the photogrammetrically acquired 3D Building Models. This dataset is derived from the City of Cape Town’s 2023 high-resolution aerial imagery and nDSM derived from the 2023 LiDAR dataset. The aerial imagery is orthorectified and has three spectral bands: visible red, visible green, and visible blue, with a spatial resolution of 8 cm. The LiDAR point cloud has a density of 10 points per square meter and a vertical accuracy of 0.1m (95%).</SPAN></P><P><SPAN>The dataset is automatically generated using trained Mask R-CNN with Precision, Recall, and F1-score of 0.85, 0.70, and 0.77, respectively; and Unet deep learning models with Precision, Recall, and F1-score of 0.97, 0.89, and 0.93, respectively. The Mask R-CNN model is trained with nDSM, and the Unet model is trained with aerial imagery. These trained models are used to extract building footprints for the entire city. The extracted building footprints showed irregularities, which were then post-processed using ArcGIS's geoprocessing model created with Esri's ModelBuilder. This geoprocessing model is run on the raw extracted footprints to produce a regularized building dataset. Each 2D building footprint has a mean building height calculated from the 2023 LiDAR nDSM.</SPAN></P><P><SPAN>This dataset will be updated periodically to align with the latest acquired LiDAR and/or aerial imagery dataset. The Low-detail footprint layer aims to supply the City with a building footprint layer that has metro-wide coverage.</SPAN></P></DIV></DIV></DIV> |
|
licenseInfo:
|
<DIV STYLE="text-align:Left;"><DIV><DIV><P><SPAN>No restrictions on the digital file for non-commercial purposes. All rights reserved. No part of this data may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage or retrieval systems, without permission from the City of Cape Town Geomatics Branch. Where the data is used as a basis for or as an insertion in a map, acknowledgment must be made to the source of such data as well as the copyright of the CCT.</SPAN></P></DIV></DIV></DIV> |
|
catalogPath:
|
|
|
title:
|
2D Building Footprints Low Details |
|
type:
|
|
|
url:
|
|
|
tags:
|
["Building","Footprint","LiDAR","Normalised Digital Surface Model (nDSM)","Aerial Imagery","2D","Low Detail","Deep Learning","Mask R-CNN","Unet"] |
|
culture:
|
en-ZA |
|
portalUrl:
|
|
|
name:
|
|
|
guid:
|
|
|
minScale:
|
150000000 |
|
spatialReference:
|
|