Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Identifying every building's function in large-scale urban areas with multi-modality remote-sensing data

Zhuohong Li, Wei He, Jiepan Li, Hongyan Zhang

TL;DR

This work tackles the challenge of mapping building footprints and per-building functions across large urban areas by leveraging multi-modality remote-sensing data and sparse labels. It proposes a semi-supervised framework based on seven-band input cubes (optical imagery, building height, nighttime lights) and AOI, building-mask weak labels, trained with a modified loss on HRNet to jointly predict footprints and seven building-function types. Evaluation in Shanghai shows high agreement with government surveys (OA ~82%, Kappa ~71%) for 1.6 million buildings and an IoU of 75% for footprints, with validation against 20,000 points and official reports. The approach enables scalable, data-driven urban management and supports sustainable urban development, with open access data and maps.

Abstract

Buildings, as fundamental man-made structures in urban environments, serve as crucial indicators for understanding various city function zones. Rapid urbanization has raised an urgent need for efficiently surveying building footprints and functions. In this study, we proposed a semi-supervised framework to identify every building's function in large-scale urban areas with multi-modality remote-sensing data. In detail, optical images, building height, and nighttime-light data are collected to describe the morphological attributes of buildings. Then, the area of interest (AOI) and building masks from the volunteered geographic information (VGI) data are collected to form sparsely labeled samples. Furthermore, the multi-modality data and weak labels are utilized to train a segmentation model with a semi-supervised strategy. Finally, results are evaluated by 20,000 validation points and statistical survey reports from the government. The evaluations reveal that the produced function maps achieve an OA of 82% and Kappa of 71% among 1,616,796 buildings in Shanghai, China. This study has the potential to support large-scale urban management and sustainable urban development. All collected data and produced maps are open access at https://github.com/LiZhuoHong/BuildingMap.

Identifying every building's function in large-scale urban areas with multi-modality remote-sensing data

TL;DR

This work tackles the challenge of mapping building footprints and per-building functions across large urban areas by leveraging multi-modality remote-sensing data and sparse labels. It proposes a semi-supervised framework based on seven-band input cubes (optical imagery, building height, nighttime lights) and AOI, building-mask weak labels, trained with a modified loss on HRNet to jointly predict footprints and seven building-function types. Evaluation in Shanghai shows high agreement with government surveys (OA ~82%, Kappa ~71%) for 1.6 million buildings and an IoU of 75% for footprints, with validation against 20,000 points and official reports. The approach enables scalable, data-driven urban management and supports sustainable urban development, with open access data and maps.

Abstract

Buildings, as fundamental man-made structures in urban environments, serve as crucial indicators for understanding various city function zones. Rapid urbanization has raised an urgent need for efficiently surveying building footprints and functions. In this study, we proposed a semi-supervised framework to identify every building's function in large-scale urban areas with multi-modality remote-sensing data. In detail, optical images, building height, and nighttime-light data are collected to describe the morphological attributes of buildings. Then, the area of interest (AOI) and building masks from the volunteered geographic information (VGI) data are collected to form sparsely labeled samples. Furthermore, the multi-modality data and weak labels are utilized to train a segmentation model with a semi-supervised strategy. Finally, results are evaluated by 20,000 validation points and statistical survey reports from the government. The evaluations reveal that the produced function maps achieve an OA of 82% and Kappa of 71% among 1,616,796 buildings in Shanghai, China. This study has the potential to support large-scale urban management and sustainable urban development. All collected data and produced maps are open access at https://github.com/LiZhuoHong/BuildingMap.
Paper Structure (8 sections, 1 equation, 7 figures, 1 table)

This paper contains 8 sections, 1 equation, 7 figures, 1 table.

Table of Contents

  1. Introduction
  2. methodology
  3. Data collection
  4. Urban building function mapping
  5. Experiments
  6. Qualitative analysis
  7. Quantitative analysis
  8. Conclusion

Figures (7)

  • Figure 1: Illustration of building function classification using multi-modality input (Source) and sparse VGI label (Guide) to generate HR building function maps (Target).
  • Figure 2: Overall workflow of the proposed framework. The framework includes three main parts: (a) Data collection; (b) Urban building function mapping; (c) Accuracy assessment. The optical images (OI), Building height (BH), and Nighttime-light data (NTL) contain three, one, and three bands, respectively.
  • Figure 3: Demonstration of the multi-modality data, the produced building function maps, and corresponding building footprints.
  • Figure 4: Demonstration of the overall results and evaluation in the study area.
  • Figure 5: Pattern analysis and street views of typical buildings in Shanghai.
  • ...and 2 more figures