Hey everyone, I’m stuck trying to get my Arducam IMX477 (12MP) working on a Raspberry Pi 3 Model B Plus Rev 1.3. I ran the install script from the docs and it keeps failing with this:
Cannot find the corresponding deb package, please send the following information to [email protected]
The error also dumped my kernel version along with the board model. I’m on a fairly fresh RaspiOS (Bullseye). Has anyone else hit this? Not really sure where to go from here.
It looks like you might have been following installation instructions that aren’t intended for the Pi 3 — that particular guide may be geared towards the Jetson platform and the script can’t locate a compatible deb package for your board.
For the Raspberry Pi Zero through Pi 3, there’s a separate set of steps for the 12MP IMX477. Try following the guide here instead:
https://docs.arducam.com/Raspberry-Pi-Camera/Native-camera/12MP-IMX477/#raspberry-pi-zero-to-pi-3
That should sort out the driver installation. Give it a shot and let us know how it goes.
That did the trick — drivers downloaded and installed without a hitch, thanks!
But now I’ve run into something else. I’m trying to access the camera through Python with OpenCV (using the standard cv2.VideoCapture with a Gstreamer pipeline) and I’m getting:
a pipeline cannot be accessed
I’ve tried this on both 64-bit Bullseye and Bookworm, clean installs on both, same error on each. The camera shows up on the i2c bus so it seems connected fine. Anyone know if this is a Gstreamer version thing or something with the Arducam driver?
That Gstreamer error is almost certainly down to the custom pipeline string you’re passing into OpenCV — things like caps negotiation, element ordering, or missing plugins can cause that “pipeline cannot be accessed” message regardless of whether the camera hardware and drivers are fine. Custom Gstreamer pipeline configuration inside OpenCV falls a bit outside what we can directly support on the Arducam side since there are so many variables.
Before diving deeper into the OpenCV pipeline, let’s first confirm the camera hardware is actually working correctly through the standard software stack. Can you test it with rpicam-hello from the libcamera utilities? That’ll tell us right away whether the driver and camera are solid:
https://docs.arducam.com/Raspberry-Pi-Camera/Native-camera/Libcamera-User-Guide/#rpicam-hello
Run that and let me know what you get — if rpicam-hello works, then we at least know the problem is isolated to the OpenCV/Gstreamer pipeline and not the camera setup itself.