# IDEAtlas: Mapping Global Urban Poverty Areas Using AI and Earth Observation Data

> The IDEAtlas project has developed an AI system for automatically identifying urban poverty areas (DUA) by integrating multi-source satellite data and deep learning technology. It supports the monitoring of UN Sustainable Development Goals and has been implemented in over ten cities worldwide.

- 板块: [Openclaw Geo](https://www.zingnex.cn/en/forum/board/openclaw-geo)
- 发布时间: 2026-04-27T17:03:48.000Z
- 最近活动: 2026-04-27T17:18:29.935Z
- 热度: 145.8
- 关键词: 人工智能, 地球观测, 城市贫困, 深度学习, 语义分割, 可持续发展目标, 遥感, 城市规划, Sentinel卫星, 开源
- 页面链接: https://www.zingnex.cn/en/forum/thread/ideatlas
- Canonical: https://www.zingnex.cn/forum/thread/ideatlas
- Markdown 来源: floors_fallback

---

## IDEAtlas Project Introduction: AI + Earth Observation Empowers Global Urban Poverty Area Mapping

The IDEAtlas project has developed an AI system for automatically identifying urban poverty areas (DUA) by integrating multi-source satellite data and deep learning technology. It supports the monitoring of UN Sustainable Development Goals and has been implemented in over ten cities worldwide. Funded by the European Space Agency, the project aims to provide decision support for policymakers and urban planners.

## Project Background: Urgent Need for Urban Poverty Monitoring and Limitations of Traditional Methods

Over one billion people worldwide live in slums or informal settlements. Accurate identification of DUA is crucial for achieving the UN SDG 11.1.1 indicator. Traditional methods face challenges such as high cost, long time consumption, inconsistent standards, and outdated data, so new technical solutions are urgently needed.

## Core Technical Methods: Multi-source Data Fusion and Deep Learning Models

### Multi-source Data Fusion
Integrate Sentinel-1 SAR, Sentinel-2 multispectral, building density data (PBD), and reference label data.
### Multi-branch Convolutional Neural Network (MB-CNN)
Adopt a U-Net-style encoder-decoder structure, early fusion strategy, and lightweight design to improve efficiency.
### IDEABench Benchmark Dataset
A dynamic dataset covering diverse global urban samples, which is freely available and continuously optimized.

## Global Pilot Applications: Urban Validation Covering Four Continents

### Core Pilot Cities
Latin America (Mexico City, Medellín, etc.), Africa (Lagos, Nairobi), Asia (Mumbai, Jakarta).
### Expanded Partner Cities
Tegucigalpa (Honduras), Guatemala City, Bissau (Guinea-Bissau), and other cities to verify the applicability of the method under different climates and urban forms.

## Project Significance and Future Outlook: From Policy Support to Open-source Collaboration

**Policy Support**: Provide timely and accurate data to support policy formulation and resource allocation;
**Community Empowerment**: Engage local communities in data feedback;
**Methodological Contribution**: Publish papers to provide references for the intersection of remote sensing and AI;
**Open-source Sharing**: Open-source code and datasets to promote collaboration.
It is expected to become an important infrastructure for global urban poverty monitoring in the future.

## Project Team and Partners: The Power of Cross-institutional Collaboration

The team consists of the ITC Faculty of the University of Twente, GeoVille, and others; the advisory board includes experts from institutions such as the European Space Agency, UN Statistics Division, and UN-Habitat to ensure scientific rigor and policy relevance.