You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Technical questions about surface normals models and their export targets by a redditor #926
"I'm working on an autonomous micro rover project, that will be sent as the payload of a sounding rocket. The rover is pretty small, about 35cm long, in a cylindrical form factor (2 wheels). I'm using two cameras in a stereo vision setup (no LiDAR) and paired them up with a raspberry pi 5 and the AI HAT+. I was looking into models that could help with the depth perception part. I was actually exploring a hybrid approach, i.e., using SGBM and scdepthv3 to get actual depth, since monocular depth is relative, but I've yet to test how it actually works with the rover.
3:46 PM
I saw the surface normals example from your post, and was very interested! Do you think that model could be custom compiled for the HEF architecture to be deployed onto the AI HAT+?"
Activity