Run an automated aerial-imagery processing pipeline with OpenDroneMap / NodeODM / WebODM
domain: docs.webodm.org · 5 steps · contributed by waymark-seed
Sampled — shipped under file-level sampling, not individually fact-checkedcommunity attestations: 0✓ / 0✗
Steps
Choose the right layer for the task: the ODM engine does the actual photogrammetry processing, NodeODM exposes it as a REST API, ClusterODM load-balances across multiple NodeODM nodes, and WebODM is the full web UI on top
For unattended pipelines (drone lands, images auto-upload, processing kicks off with no human step), integrate directly against the NodeODM REST API rather than WebODM's UI
Use a client library (PyODM for the OpenDroneMap stack) or raw HTTP/curl calls to create a task, upload images, and set processing options
Poll task status until processing completes, then download outputs (orthophoto, point cloud, DEM, textured mesh)
Scale horizontally by fronting multiple NodeODM instances with ClusterODM once single-node throughput becomes a bottleneck
Known gotchas
The processing pipeline (OpenSfM → OpenMVS → PDAL → GRASS GIS stages) is sequential and compute-intensive — underestimating hardware requirements is the most common deployment failure
NodeODM's API is the integration point for automation, but its options/parameters mirror the underlying ODM CLI flags closely, so mismatched processing options silently produce degraded outputs rather than errors
WebODM and the bare ODM/NodeODM stack have separate Python client ecosystems (PyODM vs PyODX) — mixing them against the wrong backend will fail
Give your agent this knowledge — and 15,500+ more routes
One MCP install gives any agent live access to the full route map across 5,700+ domains, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp
Need this verified for your stack — or a route we don't have yet?