How to get the bounding box coordinates(face label's in deepstream-peoplenet model-app

I am new to deepstream development.

What I did are: Using GPU, Ubuntu, run the deepstream-app-source1-peoplenet with pruned peoplenet model.

How to do:

  1. Is it possible to get the face label’s bounding box coordinates? Since I saw fairly good detection of the faces in this model.

2)How to use the plugin to do that? Can Gst-nvdsosd do this job for me?

3)To do the capturing coordinates job, deepstream app can do that already or I need to choose python app or .cpp.

Appreciate it!

Hey, nvinfer plguin will attach the bbox coordinates into meta data, so you can get them by installing a probe on any downstream plguin of nvinfer no matter c or python

Thank you! I am literally studying from zero.

Make a try right now.

Hey, Can you show me how to install a probe on the plugin? I am literally starting from zero, so please don’t mind any simple questions from me. Thanks in advance.

Hi zeeshanjafferi,

Please help to open a new topic for your issue. Thanks

@keionlu can you share the source code to run peoplenet using deepstream