Arducam 16MP IMX519 camera installation error on Jetson Orin Nano: cannot find deb package

Hi everyone, I’m trying to get an Arducam 16MP (IMX519) working on my Jetson Orin Nano, following the Native Camera Quick Start Guide at Quick-Start-Guide - Arducam Wiki. When I run the install steps I get:

Cannot find the corresponding deb package, please send the following information to [email protected]

My kernel is 5.10.104-tegra-35.3.1-20230319081403 and the board is an NVIDIA Orin Nano Developer Kit. For what it’s worth, an IMX219 works fine on this board, but the Arducam 64MP seems unsupported, the 16MP IMX519 gives that error, and I’m not sure about the Raspberry Pi HQ camera IMX477. Anyone run into this?

Based on that error and your kernel version (5.10.104-tegra-35.3.1), Arducam hasn’t released a native camera driver for the Orin Nano platform yet. That’s why the install script can’t find a matching deb package. IMX219 works because it’s covered by the stock Jetson drivers. There isn’t a driver package to install for the IMX519 on Orin Nano right now - Arducam says they’re actively working on Orin Nano support and hope to release it in the near future.

Thanks. Since the Orin Nano driver isn’t ready, I set up a regular Jetson Nano instead and tried the install there. Running ./install_full.sh -m imx519 gives the same deb package error:

Cannot find the corresponding deb package, please send the following information to [email protected]
Kernel version: 4.9.337-tegra-32.7.4-20230608212426
Jetson type: NVIDIA Jetson Nano Developer Kit.

I’m using the Arducam 16MP and 64MP cameras. I checked the supported JetPack reference at Introduction to Arducam Jetson Cameras - Arducam Wiki. Am I still on an unsupported version?

Right, your JetPack/L4T version on the Jetson Nano isn’t in Arducam’s supported list. The kernel 4.9.337-tegra-32.7.4 corresponds to L4T 32.7.4, but the latest supported version is JetPack 4.6.3 / L4T 32.7.3. You’ll need to reflash the board to a version in the supported list (4.6.3 / L4T 32.7.3) and then rerun the install script. Also heads-up: the 64MP camera currently only works on Raspberry Pi, so don’t expect it to work well on Jetson devices.

Okay, I reflashed to JetPack 4.6.3 / L4T 32.7.3 and the driver install completed without errors this time. But the Gstreamer commands from the guide’s ‘Using Gstreamer access camera’ section don’t work for me. These two fail:

gst-launch-1.0 v4l2src device=/dev/video0 ! video/x-raw, format=GAY8, width=2560, height=800 ! videoconvert ! xvimagesink

gst-launch-1.0 v4l2src device=/dev/video0 ! ‘video/x-raw,format=(string)GRAY8,width=(int)2560,height=(int)800’ ! nvvidconv ! ‘video/x-raw(memory:NVMM), format=(string)I420’ ! omxh264enc ! ‘video/x-h264, stream-format=(string)byte-stream’ ! h264parse ! qtmux ! filesink location=test.mp4 -e

The only one that works is:

gst-launch-1.0 nvarguscamerasrc ! ‘video/x-raw(memory:NVMM),width=1920,height=1080,format=NV12’ ! nvoverlaysink

I can capture an image that way, but it’s not clear/blurry. How do I control autofocus?

The v4l2src pipelines aren’t the right route for this camera - the nvarguscamerasrc pipeline is the one to use, which is why that one works. For the blurry image, you need to drive the IMX519 autofocus. There’s a focus example for exactly this: ArduCAM/Jetson_IMX519_Focus_Example on GitHub. Use that to control the lens position and focus.

I tried the focus example but it won’t open the camera. Running ‘python FocuserExample.py -i 7’ gives this traceback:

Traceback (most recent call last):
File “FocuserExample.py”, line 193, in : main()
File “FocuserExample.py”, line 181, in main: camera = Camera()
File “/home/nvidia/Downloads/Jetson_IMX519_Focus_Example-master/JetsonCamera.py”, line 107, in init: self.open_camera(width, height)
File “/home/nvidia/Downloads/Jetson_IMX519_Focus_Example-master/JetsonCamera.py”, line 112, in open_camera: raise RuntimeError(“Failed to open camera!”)
RuntimeError: Failed to open camera!

The nvarguscamerasrc command still works, but the picture is blurry.

That traceback points to open_camera() in JetsonCamera.py failing. One thing that trips people up is having opencv-related dependencies pre-installed that conflict with the example. Did you install any opencv packages before running it? Also try the other focus indices and interpreters to rule out an index/interpreter issue:

python FocuserExample.py -i 8
python3 FocuserExample.py -i 7
python3 FocuserExample.py -i 8

I tried python FocuserExample.py -i 8 and python FocuserExample.py -i 7, both fail with the same RuntimeError: Failed to open camera!:

Traceback (most recent call last):
File “FocuserExample.py”, line 193, in : main()
File “FocuserExample.py”, line 181, in main: camera = Camera()
File “/home/nvidia/Downloads/Jetson_IMX519_Focus_Example-master/JetsonCamera.py”, line 107, in init: self.open_camera(width, height)
File “/home/nvidia/Downloads/Jetson_IMX519_Focus_Example-master/JetsonCamera.py”, line 112, in open_camera: raise RuntimeError(“Failed to open camera!”)
RuntimeError: Failed to open camera!

opencv itself imports fine in both python and python3.7, so the import isn’t the problem.

The issue is likely an opencv version mismatch. The focus example expects opencv 4.5 or 4.1; if you’ve got opencv 4.8 (or another unsupported version), camera open can fail exactly like this even though the import works. The supported versions are only 4.5 and 4.1. Cleanest fix is to re-burn the OS with SDKmanager and install the relevant opencv packages:

sudo apt install nvidia-opencv
sudo apt install nvidia-opencv-dev

So the current conclusion: to get FocuserExample.py to open the camera and control autofocus, you need opencv 4.5 or 4.1 - opencv 4.8 won’t work.