Hi I am having some memory issues with the dnn_encoder node from isaac_ros_dnn_inference. A more thorough description is with the error message and code is here https://github.com/NVIDIA-ISAAC-ROS/isaac_ros_dnn_inference/issues/29

I realize that it is a problem with the Nitros node where the encoder is set up to perform pre processing, such as resizing, and this is sequestering a lot of memory.
For my project, I don’t need to do any preprocessing at all, I just need to format the input raw image into the float tensorList that is expected for the triton/tensorRT node. The image is guaranteed to be the correct size already. How would I go about modifying the encoder node to behave this way?

The conversation for this is on the GH Issue: dnn_encoder GXF_OUT_OF_MEMORY · Issue #29 · NVIDIA-ISAAC-ROS/isaac_ros_dnn_inference · GitHub