Hi everyone, I’m running two Arducam 20MP USB 3.0 camera modules (B0477, with the 16mm C-mount lenses) on an NVIDIA Jetson Orin. They’re supposed to run continuously - my script waits for an external input, then captures images. After a few captures it dies with:
VIDEOIO(V4L2:/dev/video0): select() timeout
The cameras work intermittently and it’s pretty unstable overall. I don’t think this is a bandwidth issue… anyone seen this kind of error before?
Thanks for the report. A few things would really help narrow this down. Can you share the script you’re using to run both cameras and save the images? Also, if possible, a short video showing the whole error process from start to failure would be useful. And what resolution and frame rate are you configuring for each camera? Resolution and frame rate directly affect the data throughput over USB, which could definitely tie into a select() timeout.
The flow is basically: wait for external input → capture from both cameras → save images. After a few images it fails. Though now the error has changed a bit. It prints ‘Right Piece: True’ and then:
[ WARN:[email protected]] global /io/opencv/modules/videoio/src/cap_v4l.cpp (889) open VIDEOIO(V4L2:/dev/video1): can’t open camera by index
Right Piece: True
Traceback (most recent call last):
File “/home/w28/Documents/Weld-Vision/multi-cam.py”, line 132, in
assert capR.isOpened()
AssertionError
So the first camera seems okay but the second one won’t open. Still don’t think it’s bandwidth - both cameras are on separate USB 3.0 ports.
Got the script, thanks. Looking at your log, the failure has actually shifted from the original select() timeout on /dev/video0 to the right camera not opening at all. The ‘can’t open camera by index’ warning for /dev/video1 means V4L2 couldn’t grab it, and then line 132’s assert capR.isOpened() fails right after.
So the left camera opens (that’s the ‘Right Piece: True’ print), but capR on /dev/video1 doesn’t. Need to figure out why /dev/video1 isn’t available at that point. I’ll run your script on our end to try to reproduce it - I’ll set aside some time next week to test and update you here once I have progress.
Apologies for the delay - it’s been a busy week on our side and I haven’t gotten to the bench test yet. It’s on my to-do list though. I’ll post back here as soon as I’ve run the script and dug into the /dev/video1 failure.