ROS Deep Learning CSI Camera parameters

[muhammedsezer12]

Hi,

I download ROS Deep Learning package for jetson nano and tried camera source.
But this node starts camera with mode 5 (720*1280) and this mode take image with narrow angle.
camera can not see a large angle which I need. so I need to start camera with mode 1 or 2 and take image with resolution 720*1280 with 21 fps.

For example: in opencv I can start camera object with these settings and capture image.
But i could not manage it in ros deep learning by changing parameters.
How can I do this?

Thanks,

[dusty_nv]

Hi @muhammedsezer12, I don’t explicitly set the sensor-mode in nvarguscamerasrc element from GStreamer, instead I allow it to select the best mode based on the requested resolution.

Have you tried setting the width and height parameters to the video_source node so that it selects your desired sensor mode? https://github.com/dusty-nv/ros_deep_learning#video_source-node

Alternatively, you can modify the GStreamer pipeline used in jetson-inference/utils/camera/gstCamera.cpp:144 to explicitly set your desired sensor-mode :

https://github.com/dusty-nv/jetson-utils/blob/1f3709f48258c2d75500c35605e8f6f4a3447afc/camera/gstCamera.cpp#L144

If you change the C++ code, re-run make and sudo make install after in jetson-inference/build

[muhammedsezer12]

Thanks, i will try and update this topic.