Related Experiment Video
Updated: Oct 2, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.0K
Automatic Building Extraction on Satellite Images Using Unet and ResNet50
Waleed Alsabhan1, Turky Alotaiby1
1King Abdulaziz City for Science and Technology, National Center for Data Analytics and Artificial Intelligence, P.O. Box 6086, Riyadh 11442, Saudi Arabia.
Computational Intelligence and Neuroscience
|February 28, 2022
Summary
Artificial intelligence (AI) automates building extraction from high-resolution satellite images using U-net architecture. This AI approach achieves high accuracy, aiding urban planning in dense cities.
Area of Science:
- Remote Sensing
- Urban Planning
- Computer Vision
Background:
- Rapid urbanization presents challenges in settlement planning and replanning, particularly in low-income countries.
- Manual building extraction from satellite imagery is time-consuming and labor-intensive.
- Unplanned urban settlements are a growing concern globally.
Purpose of the Study:
- To propose an automated building extraction method using artificial intelligence on high-resolution satellite images.
- To evaluate the effectiveness of the U-net architecture for image segmentation of buildings.
- To assess the potential of AI in addressing building extraction challenges in dense urban areas.
Main Methods:
- Utilized the Massachusetts building dataset, focusing on residential buildings in Boston.
- Implemented the U-net architecture with various encoders for image segmentation.
- Compared the performance of different encoder configurations within the U-net model.
Main Results:
- Achieved 82.2% Intersection over Union (IoU) accuracy for building segmentation.
- Obtained a high F1 score of 0.9, indicating precise building identification.
- Demonstrated an overall image segmentation accuracy of 90%.
Conclusions:
- Artificial intelligence, specifically the U-net architecture, shows significant potential for automated building extraction.
- High-resolution satellite imagery combined with AI can accurately map buildings in high-density urban environments.
- This technology can support efficient settlement planning and replanning processes.
