This is the main project page describing my work to develop and document an open source workflow for drone mapping and object counting using WebODM/OpenDroneMap, QGIS with Deepness plugin, DotDotGoose, and YOLOv7. This project is being conducted during the Winter of 2023/2024.
Create a repeatable recipe that a semi-technical audience can use for setting up their own internal tooling without commercial vendor dependencies.
|November 2023 (3 weeks)||* Object detection tool research and testing|
|December 2023 (2 weeks)||* Prototype workflow setup and testing|
* Automobile and bird detection
|January 2024 (3 weeks)||* Workflow documentation development|
* Testing with additional object type detection
* Custom model training and testing
* Workflow refinement
|February 2024 (4 weeks)||* Draft release target: Feb 10, 2024|
* Final refinements
Current Working Tools
- WebODM (for aerial photo processing to orthophoto)
- QGIS (for orthophoto visualization)
- QGIS Deepness Plugin (for object detection)
- DotDotGoose (for training set development)