Our products

DISHA’s portfolio currently includes two solutions that leverage AI and dynamic data for disaster response planning. 

  • The Socio-Economic Mapping solution uses mobile phone data to estimate poverty levels and population movement by region.
  • The Damage Assessment solution uses AI models to help identify damaged buildings from high-resolution satellite imagery following a natural disaster. 
  • The Shelter Mapping solution uses AI models to help extract shelter footprints in refugee settlements from high-resolution satellite imagery.
Apply for Products Access

Damage Assessment

Damage assessments are widely used in the humanitarian sector for response planning and aid allocation following natural disasters.

The Damage Assessment solution uses AI models to help identify damaged buildings from high-resolution satellite imagery. The use of assistive AI technology enables human analysts to assess damages significantly faster and cover a much broader impacted area than would be feasible with a manual-only approach. 

This product is built in partnership with Google Research and the United Nations Satellite Center (UNOSAT). It is based on Google’s SKAI model originally developed as a collaborative project of Google Research, the United Nations World Food Program (WFP) Innovation Accelerator,  and InstaDeep‘s AI for Social Good (AI4SG) team.

The solution is live in beta. To learn more, please read our blog post or email disha@unglobalpulse.org.

Socio-economic Mapping

The Socio-economic Mapping product uses mobile phone data to estimate poverty levels by region. Known as “nowcasting”, this technique helps humanitarian organizations predict areas where people might need food or cash assistance. 

Additionally, by analyzing mobile phone data, such as call patterns and top-up information, DISHA detects major population movements into or out of specific regions. This real-time information assists in quickly locating affected populations during emergencies.

This product is live in beta in the Philippines. 

To request a demo of the product, please email disha@unglobalpulse.org.

Shelter Mapping

Detailed spatial data on refugee settlements, including the location and footprint of individual shelters, is essential for site planning, service delivery, and humanitarian response. Today, this data is collected manually, either by enumerator teams on the ground or by analysts digitizing shelters from satellite imagery. Both approaches are costly, slow, and difficult to keep up to date, which means that data on many of the world’s nearly 1,000 formal refugee settlements is incomplete or outdated.

The Shelter Mapping solution uses AI models to help extract shelter footprints from high-resolution satellite imagery, with the goal of expanding coverage and shortening update cycles compared to manual mapping. The outputs are intended to support a range of operational use cases for humanitarian and development actors such as UNHCR and its partners, including building sampling frames for household surveys such as the Forced Displacement Survey, monitoring changes in settlement size and density following conflict or natural hazards, and generating baseline shelter counts in new or emerging settlements where no prior data exists.

This product is built in partnership with the Office of the United Nations High Commissioner for Refugees (UNHCR) and the United Nations Satellite Center (UNOSAT), with support from the Joint Data Center on Forced Displacement (JDC). It uses Google’s Open Building model, an AI model developed by Google Research as part of the Google Earth AI models, which detects shelters from 50cm-resolution satellite imagery. The pipeline is operated by DISHA on Google Cloud Platform, so partners such as UNHCR do not need to set up or maintain their own infrastructure.

Secondary datasets of shelter centroids and settlement boundaries produced through the pipeline will be made available publicly via UNHCR’s web services as a public good.

The solution is currently in development.

Apply for Products Access