Consumer drones have made aerial imagery collection accessible to communities, NGOs, and local governments. However, once image collections reach city scale, processing thousands of photographs into orthomosaics, point clouds, and digital surface models becomes a significant computational challenge. This problem is compounded by the limitations introduced when using lightweight consumer drones designed primarily for hobbyist use.

This talk explores how the open-source project ScaleODM enables scalable processing of large drone datasets by orchestrating OpenDroneMap workloads across Kubernetes clusters. By distributing photogrammetry tasks across multiple nodes, ScaleODM allows teams to process large mapping projects using commodity infrastructure while maintaining fully open workflows.

Using a real-world example of city-scale drone imagery, this session will demonstrate:
- Challenges encountered when processing very large drone datasets
- How ScaleODM structures distributed photogrammetry pipelines
- Deploying Kubernetes-native processing for OpenDroneMap
- Performance considerations and cluster scaling strategies
- Specific challenges introduced when processing imagery from lightweight drones such as the DJI Mini series
- Lessons learned from running large-scale drone processing in practice

The presentation will also discuss how these tools enable humanitarian mapping, urban analysis, and community mapping projects by reducing the infrastructure barriers traditionally associated with large photogrammetry workloads.

Attendees will gain a practical understanding of how open-source geospatial tools can scale from small drone surveys to city-wide imagery processing while keeping infrastructure costs accessible.