Smarter Mapping for Recovery in Nepal
Disaster Response • Nepal
Nepal faces increasing floods, landslides, droughts, and heatwaves. This project is piloting a GeoAI tool built by EPFL's VITA Lab and NAXA to speed up post-disaster building damage assessment. HOT's role is project coordination and volunteer annotation support, working toward a pilot in three flood-prone municipalities in western Nepal.
The Context
While important national initiatives have advanced the use of data and technology for disaster preparedness, the post-disaster response phase in Nepal still largely relies on manual inspections, paper-based processes, and fragmented tools and datasets. This slows recovery planning and limits the timely, reliable damage intelligence needed to guide relief, rehabilitation, and reconstruction.
Nepal faces increasing climate hazards, including:
- Floods and landslides threatening livelihoods and infrastructure
- Droughts and heatwaves impacting food systems and ecosystems
- Fragmented damage assessment tools that delay recovery planning
Our Approach
EPFL's VITA Lab and ENAC, with NAXA, have built a GeoAI tool that uses object detection and change detection to identify and classify structural damage from drone imagery. The tool is being localized to Nepal's building types and environmental conditions and is hosted on EPFL's platform.
HOT's role in the project is project management and coordinating the volunteer annotators who flag and name the specific annotation task, e.g., labeling building outlines or damage classes in the imagery pipeline.
This year is a prototyping and piloting phase. The tool is not yet in full operational use.
Impact & Deployment (planned)
Once piloted, the tool is designed to reach three flood-prone municipalities in western Nepal, covering an estimated 30,000 households (around 180,000 people). Early testing suggests the tool could cut assessment time by up to 30%, though this has not yet been confirmed in the field.
If the pilot succeeds, the project aims to:
- Strengthen local capacity for post-disaster assessment
- Improve the speed and quality of damage assessment
- Support more timely and accountable decision-making for relief and reconstruction
- Build a GeoAI model that could scale to wider use across Nepal
Tools
- Special developed annotation tool by EPFL - Swiss Federal Technology Institute of Lausanne - School of Architecture, Civil and Environmental Engineering (ENAC), IT4Research team. The tool handles a pipeline of ~13k high resolution raw images from drones post disasters in Nepal and is hosted on EPFL
Mapper can log in using their email, name and code below:
Annotators’ code: nepal-2026
