How to load a model into graph composer?

Accelerated Computing Intelligent Video Analytics DeepStream SDK
,
1

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) GPU
• DeepStream Version 7.0
• JetPack Version (valid for Jetson only)
• TensorRT Version 8.5
• NVIDIA GPU Driver Version (valid for GPU only)
• Issue Type( questions, new requirements, bugs) question

I have trained a resnet on my data and have a .pb file with its weights. I want to know if there is a way to use that model on graph composer in the same way I can use the sample models used on deepstream tests 1 and 2. I know that I probably have to convert it to onnx format.
I’ve already seen similar questions on this forum but I still don’t have any idea of how I can do that. Most of them were referring to the peoplenet TAO model or something similar.

3

The onnx model can be supported.

4

What files do I have to export and where?What do I have to do in order to see the new model in graph composer just like I can see the sample models?I want to see my trained model in there.

image
image714×388 10.9 KB

5

You can use the NvDsInferVideo extension. NvDsInferenceExt — DeepStream documentation 6.4 documentation
Set the correct nvinfer configuration file in the red circle.

gxf_infer.PNG
gxf_infer.PNG1920×1095 147 KB

6

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