Orthophoto Segmented Images

ODM
#1

I’ve used ODM a lot and I’ve been really impressed with the results, even with underwater drone imagery.
I am now segmenting (with the AI) the pictures so that I can extract the classes of interest. So far having the segmented, georeferenced images doesn’t work for ODM, but my theory is that if I add points of interest on the segmented images, the algorithm will be able to identify these across images and stitch them together… Could this be possible? If yes, what kind of points of interest does the algorithm look for? Could it simple by points? or lines?

Thank you in advance for any help!

1 Like
#2

Welcome, Elisa!

That sounds like it’d be a similar implementation to Manual Tie Points, which we don’t have a concept of quite yet.

As for the default feature extraction algorithm, we’re using SIFT and/or HAHOG:
https://docs.opencv.org/master/da/df5/tutorial_py_sift_intro.html

#3

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.