I’m brand new to the project and trying to get the scope of the project. I have the docker WebODM running and am processing the auckerman data processing right now
I’m currently working on my capstone project for uni - which focuses on using TensorFlow for image classification. I’d love to contribute automated classification to the project so: I’d appreciate some advice on what has already been done and which parts of the project I should investigate.
The project looks great so far!
Hey @Samwise that sounds like a cool project!
The only classification we currently do is a statistical method to differentiate between ground/non-ground points in the point cloud output (no machine learning at this point).
If you want to implement something with Tensorflow, you’d probably want to integrate this as a plugin. We’re still in the process of building the plugin system, so if a particular hook does not exist, I’d be happy to consider adding new ones (a hook is simply an event).
See https://github.com/OpenDroneMap/WebODM/blob/volume/app/plugins/plugin_base.py and https://github.com/OpenDroneMap/WebODM/tree/volume/plugins/test
For some examples. You probably want to work off the
volume branch for now since that has the latest changes.
Thanks @pierotofy , I might hack something together using the “download assets” button to start.
I noticed that there’s an issue with the point cloud viewer that @Andes_Consulting_Eng brought up. I checked out the csv file and it seems fine, so I assume it’s an issue with the viewer - yea?
It’s a precision issue. I’ve opened a PR that solves the problem, but hasn’t been merged yet. https://github.com/OpenDroneMap/OpenDroneMap/pull/791