OpenDroneMap/ODM
ODM
OpenDroneMap (ODM) is an open-source aerial drone photogrammetry pipeline that converts 2D images into georeferenced 3D models, orthophotos, point clouds, and digital elevation maps. At its core, the OpenDroneMap Processing Engine orchestrates a complete Structure-from-Motion workflow, from feature extraction through dense reconstruction and tiled output generation, purpose-built for transforming drone-captured imagery into geospatial data products.
The toolkit distinguishes itself through GPU-accelerated SIFT feature extraction using CUDA-capable NVIDIA graphics cards, roughly doubling processing speed compared to CPU-only operation. It supports multiple dense reconstruction backends, integrates with OpenSFM for camera pose computation, and offers GeoJSON boundary restriction to limit processing to user-defined polygons. The pipeline generates OGC 3D Tiles for web-based visualization, Cloud-Optimized GeoTIFF orthophotos, and digital surface and terrain models through progressive morphological filtering.
ODM provides a REST API with a task queue for remote job submission and asynchronous result retrieval, alongside configurable feature extraction quality settings that balance memory usage, processing time, and accuracy. It handles video frame extraction with GPS subtitle integration, ground control point georeferencing, and reconstruction alignment to reference files. Additional capabilities include AI-based background removal, automatic reconstruction boundary cropping, and orthophoto generation that skips dense 3D model creation.
Features
- Structure From Motion - Orchestrates a complete multi-stage photogrammetry workflow that reconstructs 3D geometry from unordered 2D image collections.
- Aerial Photogrammetry Pipelines - Converts drone-captured 2D images into georeferenced 3D models, orthophotos, and elevation maps.
- SIFT Feature Extractors - Ships a CUDA-accelerated SIFT extractor that roughly doubles processing speed compared to CPU-only operation.
- Digital Terrain Model Generation - Builds a digital terrain model representing only the ground using a morphological filter from aerial imagery.
- Cloud-Optimized GeoTIFF Generators - Produces Cloud-Optimized GeoTIFF orthophotos and digital surface/terrain models from aerial imagery.
- Digital Surface Models - Builds a digital surface model representing ground plus objects using a progressive morphological filter.
- Orthophoto and Elevation Model Generators - Produces orthophotos, digital surface models, and digital terrain models from drone imagery.
- Photogrammetry Orchestrators - Core command-line toolkit orchestrating photogrammetry tasks from feature extraction to textured mesh generation.
- 3D Reconstruction Pipelines - Converts sets of 2D aerial photographs into accurate 3D digital models via Structure-from-Motion.
- SIFT Feature Accelerators - Extracts SIFT features on CUDA-capable NVIDIA GPUs, processing images roughly twice as fast as CPU.
- Multi-Backend Dense Reconstruction - Supports multiple backends for generating dense point clouds and meshes, including MVS-based and depth-map fusion approaches.
- Orthophoto Tiling - Divides the reconstruction area into manageable tiles for parallel processing and memory-efficient orthophoto compositing.
- Task Queues - Exposes processing jobs through a RESTful HTTP interface that queues tasks and returns results asynchronously.
- Cloud-Optimized GeoTIFF Accessors - Creates Cloud-Optimized GeoTIFF files for faster remote access and streaming of orthophoto data.
- Boundary Clipping Tools - Restricts reconstruction to a user-defined GeoJSON polygon by clipping point clouds and meshes against the boundary.
- 3D Tiles Streaming - Generates OGC 3D Tiles from the reconstruction for streaming in web-based 3D viewers.
- OGC 3D Tiles Generators - Generates OGC 3D Tiles from the reconstruction for streaming and visualization in web-based 3D viewers.
- Ground Control Point Georeferencers - Aligns aerial images using surveyed ground markers to improve the positional accuracy of the final outputs.
- Progressive Morphological Filters - Separates ground from non-ground points using a multi-scale morphological filter to produce digital terrain and surface models.
- Sparse Orthophoto Generators - Produces orthophotos directly from the sparse reconstruction, skipping dense 3D model generation for faster results.
- Video Frame Capture - Extracts individual frames from video files for subsequent photogrammetric processing.
- Geospatial Model Alignments - Aligns output models to geospatial reference files for accurate positioning in mapping workflows.
- Photogrammetry API Services - Provides a REST API with a task queue for remote photogrammetry job submission and asynchronous result retrieval.
- Photogrammetry Software - Command line toolkit for processing aerial drone imagery.
- Satellite Datasets - Toolkit for generating maps and 3D models from drone imagery.
