Imaging formats: 3D models (.obj)

[StephenChan]

Suggested by R.F in the Google Group, although this has probably come up several times overall.

Obviously, this has all sorts of implications for how CoralNet is structured, not the least which is that the term "Image" isn't general enough anymore. At a certain point we should ask ourselves: is CoralNet the be-all end-all for all possible coral analysis, or should we focus on doing one category of analysis and doing it well?

[StephenChan]

As far as 3D goes, I'm pretty sure orthomosaics specifically are what most users are looking for right now. So this goes to low priority, in favor of #192.

[beijbom]

Full support for orthos is a whole new research project from the perspective of the automated annotation side.

On Wed, Jun 27, 2018 at 12:52 AM StephenChan notifications@github.com wrote:

As far as 3D goes, I'm pretty sure orthomosaics specifically are what most users are looking for. So this goes to low priority, in favor of #192 https://github.com/beijbom/coralnet/issues/192.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/beijbom/coralnet/issues/169#issuecomment-400577502, or mute the thread https://github.com/notifications/unsubscribe-auth/ABE5l7RqxUFK4l8ILOjmENCAMXr3kgIeks5uAznXgaJpZM4UfNEU .

[StephenChan]

Full support for orthos is a whole new research project from the perspective of the automated annotation side.

That makes sense, although doing automated annotation based on a top-down 2D snapshot of the ortho might be good enough in the near term. If orthomosaics speed up manual processing, data collection, etc. such that the backend has more data to train on, then that's going to improve the backend results already.

[beijbom]

True enough.