Mapping Deprived Urban Areas using EO & AI
The rapid urbanization and population growth across many regions worldwide have significantly increased the spread of deprived urban areas (DUA), often called slums or informal settlements. The lack of reliable geospatial information on their extent and locations in many cities continues to hinder efforts aimed at improving living conditions. The IDEAtlas initiative addresses this critical information gap by employing a User- and Data-centric Artificial Intelligence (AI) approach for accurately mapping these areas to support Sustainable Development Goal (SDG) Indicator 11.1.1. This workshop equips you with the technical foundation to implement and contribute to this effort.
Workshop Objectives
The objective of this training is to equip participants with the skills to implement an end-to-end mapping pipeline. By the end of this workshop, you will be able to:
1. Data Creation & Annotation
- IDEAtlas User Data Portal: Utilize the IDEAtlas User Data Portal to digitize polygons and generate the reference data required for training the AI model.
2. The Mapping Workflow
You will master the full lifecycle of a DUA mapping project, including:
- Classification: Execute inference using the IDEAtlas pre-trained models to map a city’s DUA extent.
- Fine-tuning: Adapt existing pre-trained models to new geographic contexts when local reference data is limited.
- Training: Train custom models from scratch for specific city where sufficient training data is available.
- Indicator Derivation: Process classified maps to generate summary statistics aligned with SDG 11.1.1.