i have generated a map of informal settlement and there are so many houses to count but it seems like DroneDeploy wants you to count them manually, my question is that, is there another way to automatically count structures rather than counting manually by hand?
2 Likes
I assume when you say “manually by hand” you mean with the count annotation? I know the system has been trained for things like solar panels but because house can vary so wildly in their appearance from the air it’s probably unlikely. It shouldn’t take very long at all to tag a couple hundred houses with the annotation tool.
2 Likes
so, there is no other way of counting structures besides counting them munnually?
1 Like
@bongani to caveat on @MichaelL comment. Currently, the count function will not identify houses due to the reasons previously stated. I would say this is a great feature request and would be appropriate to post in the DroneDeploy forum for future ideas and functionality.
1 Like
@Rob_Corbett thank you for your response, i have been searching for the forum and i can find it can you please provide me with a link for it if you have.
1 Like
I have modified the thread to Feature Request which will also help it stay on the board. Hopefully it will generate some similar use cases and suggestions. I agree it would be very helpful on tasks like this. I am finishing up an existing conditions/topo with a local RPLS of a 3 mile roadway for Williamson county. There were 72 structures and it took me about 45-60 minutes to tag them. The time consuming part was labeling each one. I don’t know if this is a feasible task but there has to be a better way to identify markers without clicking on each one and opening another panel to see each one. A little different than the initial part you are looking for, more of an extension of the function.
1 Like
As part of a feature, have a teach/learn mode of what to look for and count. For example, outline a couple of subjects (e.g. houses, RTUs, GCPs, cars, trees, etc), the functionality then shows you what it things are the objects, then feedback loop to tighten up what to look for, and then provide count.
1 Like
Can’t really see how you’d pull those biometrics out of photogrammetry data; it’s all photos from above canopy level. Theres all sorts of crazy stuff you can do with multispectral imaging that allows you to determine species and health characteristics from particular signatures.
As others have said, you’ll likely need a Lidar unit to penetrate the canopy. But I think you’d still have to fly lots of overlapping passes at oblique angles to try and capture that info
1 Like