- Edited
Hey guys, I'm back
(or rather I saw the announcement, that I should write here on the forums now instead on discord)
This one is again a bit difficult to describe/put in in words, but I'll try anyway.
We're basically in the stage to make the issue at least reproducible.
So my colleague has noticed some odd behaviour:
Usually the cams work all fine with the depth estimation,
but in some cases, the estimation is way off (like 30% or so).
Difficult to have stats right now, he can only describe it currently by the Anchorman meme "60% of the time, it works every time"
It seems after starting it's either working fine all the time,
or works "ok" all the time,
or the measurements are way off all the time.
So we suspect some calibration issue after starting the script.
(I will try to get it and post it here too, when I have it)
current infos that I got so far:
- dai library versions are the latest
- happens on win & linux
- tested on 2 different cams (oak-d-s2)
So it'd be interesting, if anyone else encoutered anything similar (I haven't seen any recent posts with anything like this)
and would be nice, if you guys could provide some suggestions/ideas how we should debug this issue/ what debug data to collect, etc ...
(I'm thinking of some setup with a fixed position of the cam,
we trigger the script, as long the output fits the expected values, we just shutdown and restart it,
and in case the measurements are off, we collect debug data)