As we close in on good improvements to the experience for users and developers on OpenDroneMap, including, thanks to Michael a 3.6.1 release:
the new Pixi build system (in progress and nearing completion):
and now Leonardo’s OpenSfM upgrades (in progress):
we are better poised than ever to properly build out and sustain OpenDroneMap.
There’s been one notable gap. While oats is a good testing framework, we don’t have reference data for properly discerning whether changes to the processing pipeline improve or degrade the quality of outputs.
To that end, @huey and I have been working on some fairly promising looking synthetic datasets. For now, I’ll just share a teaser, with more updates and details shortly. This shows the DSM directly from the synthetic data (super detailed with individual leaves of a tree), and the output of the DSM as processed through OpenDroneMap.
This is a very preliminary comparison, but a nice teaser of what is to come for better testing datasets.