Retain the trt model on GPU after processing the entire video

Accelerated Computing Intelligent Video Analytics DeepStream SDK
, ,
1

• Hardware Platform (Jetson / GPU): GPU
• DeepStream Version: 6.3
• TensorRT Version: 8.5.3-1+cuda11.8
• NVIDIA GPU Driver Version (valid for GPU only): 525.89.02
• Issue Type( questions, new requirements, bugs): questions

How can’t i retain the trt model on GPU after processing the entire video in the application and add the new video to run, i must not load engine again.
I try runtime_source_add_delete but when it warning:

Warning: gst-resource-error-quark: No Sources found at the input of muxer. Waiting for sources. (3): gstnvstreammux.cpp(2777): gst_nvstreammux_src_push_loop (): /GstPipeline:pipeline0/GstNvStreamMux:Stream-muxer

I can’t add source again. Thank you

3

I don’t quite understand your needs. Are you having problems using runtime_source_add_delete?

If you want to add and remove sources dynamically, consider using nvmultiurisrcbin

https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_plugin_gst-nvmultiurisrcbin.html

4

Can you give me an example about nvmutilurisrcbin using python. Thank you

5

Just use it like a normal element. When you need to add or remove a source, send the corresponding http request. The parameters required by these requests are described in detail in the documentation.

    source = Gst.ElementFactory.make("nvmultiurisrcbin", "nvmultiurisrcbin")
    if not source:
        sys.stderr.write(" Unable to create Source \n")

/opt/nvidia/deepstream/deepstream/service-maker/sources/apps/python/pipeline_api/deepstream_test5_app/deepstream_test5.py