Customize the preprocess plugin of nvinferserver

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Please provide complete information as applicable to your setup.

**• Hardware Platform (GPU) T4
**• DeepStream Version 6.1.1 triton container
**• Issue Type( questions)
Nvinferserver guide manual deatails : The Gst-nvinferserver Plugin’s default preprocessing cannot be disabled. Color conversion, datatype conversion, input scaling and object cropping are continue working in nvds_infer_server natively. The Gst-nvinferserver plugin can support Triton ensemble models for further custom preprocessing, backend and postprocessing through Triton custom-backends.
My questions Details : DS’s built-in preprocess plugin doesn’t match the preprocess of yolov5-6.0 such as padding method. The different preprocess method cause output of DS’s triton-grpc different from the out of origin mode code.
My questions : Can I customize the preprocess plugin ? How to do it ? I read the official documentation but still don’t understand，help please

3

currently the latest deepstream6.1.1 dose not support nvpreprocess + nvinferserver, only nvpreprocess + nvinfer. please refer to Gst-nvinferserver — DeepStream 6.1.1 Release documentation. you can find symmetric_padding in this link.

4

Thanks for your reply. Will nvinferserver source code be opened later? when ?

5

Can I change the padding value ? such as 114 instead of the default value 0

6
  1. about open source nvinfserver, could you share your requirement?
  2. currently, nvinferserver dose not support changing padding value, what is your use case about changing padding value?
7

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

8

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