Multiple PGIE+Tracking same pipeline

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Hi everyone

• Hardware Platform (Jetson / GPU): RTX2080TI
• DeepStream Version: 6.2
• TensorRT Version: 8.5.2-1+cuda11.8
• NVIDIA GPU Driver Version (valid for GPU only): 530.30.02
• Issue Type( questions, new requirements, bugs): Question

I am trying to integrate different pgies with their corresponding tracker in the same deepstream pipeline. Attached is the pipeline diagram of my solution:

graph.pdf (31.8 KB)

As you can see I am using multiple video inputs via RTSP (uridecodebin sources).

Continuing the discussion from Nvtracker mistake A obj_meta->s track_id for B when two pgie in front of nvtracker: it seems that the way I do it is how it is recommended by NVIDIA. However, when it comes to retrieving the metadata information I think I am having a problem. And that is that while, I include in each PGIE a gie-unique-id so that I can differentiate in the object_meta via the unique_component_id, I receive for all objects the same unique_component_id. I have to say that I add the metadata buffer treatment to the last tracker of the pipeline.

element_src_pad = last_tracker.get_static_pad("src")
    if not element_src_pad:
        loggers['error'].error("Unable to get src pad")
    else:
        element_src_pad.add_probe(Gst.PadProbeType.BUFFER, callback, 0).

Also, I see that the same object is tracked multiple times, how can I fix it?

Screenshot from 2023-06-13 21-40-50
Screenshot from 2023-06-13 21-40-501088×879 33.2 KB

Best regards

3

Hello,

Sorry for the insistence. Could someone help me?

Best regards

4

Hello,
I have the same issue, have you managed to solve it?

5

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks

Do you have the issue if only use one nvtracker in your pipeline?