Deepstream custom plugin

can you please let me know how can i edit the deepstream buffer image.
Like in custom plugin you have done blur_objects i like to apply segments on the image so want to know how can i convert dsexample image to cv2 and can apply cv2 transformation and then put back into the deepstream buffer to visualize.
i am working on this folder /opt/nvidia/deepstream/deepstream-6.0/sources/gst-plugins/gst-dsexample

In blur_objects() you got in_mat, you can edit it whatever you want

The blur object is not working for me. Can you please let me know what are necessary changes to run blur function?

Did you seted parameters full-rame=0 and blur_objects=True?

Yes i tried to set parameters in makefile also in deepstream test py file.
But i ended up getting error for mapping of 0 index in surface.
It was for memtype of cuda i think cv2 support cpu and my image in cuda so it didn’t work.
I tried to change memtype to default and got segment fault error.

can you just please let me know which deepstream_python_app i have to run for dsexample plugin?

dsexample plugin is c/c++. If you want to implement any function inside the plugin, you need to develop in c/c++.

How to execute the plugin i have to run some deepstream_python_app file, right?
Or there is any other ways?

Yes. DeepStream SDK is based on gstreamer. Gstreamer support python APIs with gst-python. DeepStream has its own python bindings, please refer to NVIDIA-AI-IOT/deepstream_python_apps: DeepStream SDK Python bindings and sample applications (

Python GStreamer Tutorial ( The plugin is just an ordinary GST element with gst-python bindings.

Please study by yourself.

1 Like

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.