Custom model in DeepStream 5.1

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU)
• DeepStream Version
• JetPack Version (valid for Jetson only)
• TensorRT Version
• NVIDIA GPU Driver Version (valid for GPU only)
• Issue Type( questions, new requirements, bugs)
• How to reproduce the issue ? (This is for bugs. Including which sample app is using, the configuration files content, the command line used and other details for reproducing)
• Requirement details( This is for new requirement. Including the module name-for which plugin or for which sample application, the function description)

Hi, I have a question about custom model in DeepStream. Is it possible in DeepStream to use TensorRT engine which is not detector, classifier or does not segmentation? I have the engine which takes as inputs frame and mask and return frame with blurred objects based on mask. Is there any option to use it or maybe should I write custom plugin for it?
Thanks

3

Currently nvinfer cannot support multiple input model, you can use videotemplate plugin to do that.

4

Thank you for reply. I implemented OpenCV bluring code in custom library in videotemplate plugin and everything works fine. So now I want to run my bluring engine inside the custom library, cause opencv code works really slow. Can you give me some instruction or template what I should implement to run tensorRT engine? Gst-nvinfer plugin is an extensive plugin and I believe I can implement code for my engine in shorter form.
Thank you

5

Can you refer deepstream_tao_apps/apps/tlt_others/deepstream-emotion-app at release/tlt3.0 · NVIDIA-AI-IOT/deepstream_tao_apps · GitHub Gst-nvdsvideotemplate — DeepStream 6.1.1 Release documentation

6

I implemented code for run engine in custom library but I get en error when I run inference. Can you help me find the reason of this error? I send my script and ONNX file in PM. Error:

…/rtSafe/cuda/genericReformat.cu (1294) - Cuda Error in executeMemcpy: 1 (invalid argument)
FAILED_EXECUTION: std::exception
free(): invalid pointer
Aborted (core dumped)

Hardware Platform: Tesla T4
DeepStream Version: 5.1
TensorRT Version: 7.2.1.6
NVIDIA GPU Driver Version: 460.73.01

My engine works with dynamic axis. It has two input images and one output images. Shape of image: ‘frame’: -1, -1, 3, ‘mask’: -1, -1, ‘output’: -1, -1, 3. To create engine:

trtexec --onnx=blend_blur_2inputs_up_to_1280.onnx --explicitBatch --fp16 --workspace=1024 --minShapes=image:220x220x3,mask:220x220 --optShapes=image:1280x1280x3,mask:1280x1280 --maxShapes=image:1280x1280x3,mask:1280x1280 --buildOnly --saveEngine=blendblur.engine

Thanks

7

Could you create a new topic for the new issue, we would like one topic to track one issue

9

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.