Description
Drone surveying software has never been shy about its price: a Pix4D or Agisoft licence runs into thousands, and a small team that uses it for a few days per project has no way to make that back.
OpenDroneMap is the open-source answer. Point it at a folder of drone photos and it returns orthophotos, digital elevation models, classified point clouds and textured 3D models — the same deliverables the commercial packages produce. It runs a mature structure-from-motion pipeline, and it is just as happy with imagery shot from a balloon or a kite.
The easiest route is Docker: `docker pull opendronemap/odm`, drop the photos into a folder named images, run one command and walk away. If you want control, the parameter list is long — DSM generation, orthophoto resolution, point cloud density and ground control points are all separately adjustable — and a dedicated GPU image makes a real difference on NVIDIA hardware. It is a command line tool by design, which suits scripting and batch work; if you would rather not type commands, there is an official OpenDroneMap Desktop interface and the WebODM browser front end.
Outputs are industry-standard formats, so nothing needs converting: GeoTIFF orthophotos and DEMs open in QGIS, LAZ point clouds in CloudCompare, OBJ and PLY meshes in MeshLab. One gotcha: those .tif files are GeoTIFFs, and Photoshop or GIMP will not open them properly — use QGIS. AGPL-3.0, 6.4k stars, Windows, macOS and Linux.
Four deliverables in one run: Orthophoto, digital elevation model, classified point cloud and textured 3D model come out of a single processing pass.
One Docker command: Pull the official image, put the photos in an images folder and start; no Python environment or dependency chain to assemble. A separate GPU image adds CUDA acceleration.
Deep parameter control: Hundreds of flags cover feature matching, dense reconstruction, mesh and orthophoto resolution and ground control, so a run can be tuned to the accuracy and hardware at hand.
Ground control points: Supply a GCP file for high-accuracy correction, an order of magnitude better in absolute position than onboard GPS alone.
Standard output formats: GeoTIFF, LAZ, OBJ and PLY open directly in QGIS, CloudCompare and MeshLab with no conversion step.
Network processing: NodeODM exposes processing over a network, with distributed options for large datasets across several machines.
Not only drone imagery: Balloon, kite and other low-cost aerial imagery works the same way, which suits community mapping and heritage recording.
Graphical front ends: OpenDroneMap Desktop provides an official GUI, and WebODM offers a browser interface for submitting jobs and reviewing results.
OpenDroneMap is the open-source answer. Point it at a folder of drone photos and it returns orthophotos, digital elevation models, classified point clouds and textured 3D models — the same deliverables the commercial packages produce. It runs a mature structure-from-motion pipeline, and it is just as happy with imagery shot from a balloon or a kite.
The easiest route is Docker: `docker pull opendronemap/odm`, drop the photos into a folder named images, run one command and walk away. If you want control, the parameter list is long — DSM generation, orthophoto resolution, point cloud density and ground control points are all separately adjustable — and a dedicated GPU image makes a real difference on NVIDIA hardware. It is a command line tool by design, which suits scripting and batch work; if you would rather not type commands, there is an official OpenDroneMap Desktop interface and the WebODM browser front end.
Outputs are industry-standard formats, so nothing needs converting: GeoTIFF orthophotos and DEMs open in QGIS, LAZ point clouds in CloudCompare, OBJ and PLY meshes in MeshLab. One gotcha: those .tif files are GeoTIFFs, and Photoshop or GIMP will not open them properly — use QGIS. AGPL-3.0, 6.4k stars, Windows, macOS and Linux.
Features
Four deliverables in one run: Orthophoto, digital elevation model, classified point cloud and textured 3D model come out of a single processing pass.
One Docker command: Pull the official image, put the photos in an images folder and start; no Python environment or dependency chain to assemble. A separate GPU image adds CUDA acceleration.
Deep parameter control: Hundreds of flags cover feature matching, dense reconstruction, mesh and orthophoto resolution and ground control, so a run can be tuned to the accuracy and hardware at hand.
Ground control points: Supply a GCP file for high-accuracy correction, an order of magnitude better in absolute position than onboard GPS alone.
Standard output formats: GeoTIFF, LAZ, OBJ and PLY open directly in QGIS, CloudCompare and MeshLab with no conversion step.
Network processing: NodeODM exposes processing over a network, with distributed options for large datasets across several machines.
Not only drone imagery: Balloon, kite and other low-cost aerial imagery works the same way, which suits community mapping and heritage recording.
Graphical front ends: OpenDroneMap Desktop provides an official GUI, and WebODM offers a browser interface for submitting jobs and reviewing results.
