Issue with DetectNet Node Processing OAK-D Camera Stream

Issue with DetectNet Node Processing OAK-D Camera Stream

Hi,

I am new to working with Isaac ROS DetectNet and am trying to integrate my OAK-D camera (using DepthAI) with the Isaac ROS DetectNet package. However, I am encountering an issue where the /detectnet_processed_image topic remains empty. My camera is streaming to the /image_rect topic, but the DetectNet node doesn’t seem to process it correctly.

Here’s a summary of my setup and what I’ve tried so far:

• The camera is streaming data to the /image_rect topic.
• I’ve verified the topic is being published correctly using rostopic list and rostopic info.
• I’ve tried adjusting the input topic in the DetectNet node to /image_rect, but the processed image topic remains empty.

I am trying to run the following commands:

1. Launch the DetectNet node:

   ros2 launch isaac_ros_examples isaac_ros_examples.launch.py launch_fragments:=detectnet interface_specs_file:=${ISAAC_ROS_WS}/isaac_ros_assets/isaac_ros_detectnet/quickstart_interface_specs.json
   

2. In a second terminal, run the dev script and setup the workspace:

   cd ${ISAAC_ROS_WS}/src/isaac_ros_common && ./scripts/run_dev.sh
   cd ${ISAAC_ROS_WS} && source install/setup.bash && ros2 run isaac_ros_detectnet isaac_ros_detectnet_visualizer.py --ros-args --remap image:=detectnet_encoder/resize/image
   

3. In a third terminal, run the dev script and open the image viewer:

   cd ${ISAAC_ROS_WS}/src/isaac_ros_common && ./scripts/run_dev.sh
   ros2 run rqt_image_view rqt_image_view /detectnet_processed_image
   

I have built the isaac_ros_detectnet package from source in every terminal, but the /detectnet_processed_image topic remains empty.

Could you please guide me through the steps to configure the input topic properly or identify any issues in my setup? Any advice or suggestions would be greatly appreciated.

Thank you for your help!

Start with isaac_ros_examples.launch.py and quickstart_interface_specs.json, then review the isaac_ros_detectnet_visualizer.py entry point and the reported /image_rect and /detectnet_processed_image topics. Reproduce the three-terminal commands and verify that the configured input produces a non-empty processed-image topic.