detectNet and custom bounding boxes in python

Robotics & Edge Computing Jetson & Embedded Systems Jetson Nano
1

Hello,
I’m working on a project where I have to chose the color of bounding boxes drawn by detectNet according to the depth from a intel realsense camera (from red to green). At the moment I’m using the bounding box coordinates from the detection object than using OpenCV I can draw boxes and stream, frame by frame, the resulting ovelayed images. But openCV halves my fps with respect of glDisplay. Is there a way to still use glDisplay with custom boxes, or an easy way to do this in python?
thanks

2

Hi mattilsanguinario, currently the detectNet bounding box colors are generated programatically by class ID in this function:
https://github.com/dusty-nv/jetson-inference/blob/8846dcb274ff812880ef7bf9d42eb1dd75c9e7b7/c/detectNet.cpp#L426

Then at runtime, the color of the bounding box is looked up by the class ID here:
https://github.com/dusty-nv/jetson-inference/blob/8846dcb274ff812880ef7bf9d42eb1dd75c9e7b7/c/detectNet.cu#L109

You would need to update these to lookup a depth image and generate the color from that, and modify the Python bindings accordingly.

3

thanks for your reply
my dream was avoid c++ because I’ve no idea how to program in c++ but I think the time to start learning it has come ;)