How to use engine inferring the multi-frame

Please provide complete information as applicable to your setup.

• Hardware Platform (GPU)
**• DeepStream Version6.1
*• TensorRT Version 8.4EG
**• NVIDIA GPU Driver Version (valid for GPU only)510
• Issue Type( questions, new requirements, bugs)

I meet a problem. please give me advice.
Here, I need to composite the 100 camera frames as a current frame, then detect the object in the composite image. How could I do that? is the preprocessing in the “gst-nvdspreprocess”?
Thank you very much.

  1. please refer to deepstream sample deeeptream-app, configuration file is /opt/nvidia/deepstream/deepstream/samples/configs/deepstream-app/source30_1080p_dec_preprocess_infer-resnet_tiled_display_int8.txt, it can do 30 video sources inference .
  2. please specify more your use case. what is media pipeline? what type is your video source?
  3. if no gst-nvdspreprocess, preprocessing is in nvifner, if have gst-nvdspreprocess, preprocessing is in gst-nvdspreprocess, please refer to Gst-nvdspreprocess (Alpha) — DeepStream 6.1 Release documentation

Thank you for your help. And I am sorry that I could not state that clearly.
My task is based on one rtsp source, and I need to composite100 frames of this one source into a picture. than we could detect the objects.
Thank you very much.

  1. if want do inference to every picture, please refer to deepstream sample deepstream-appsrc-test.
  2. if want to mux 100 frame to one picture, please refer to nvstreammux plugin, need to change nvstreammux 's batch-size and timeout, and change nvinfer 's batch-size to 100.

Thank you very much. I will try.

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