Custom sgie model is not working in graph-composer

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

I am tryng to use custom model as sgie in graph-composer

• Hardware Platform (Jetson / GPU) - GPU
• DeepStream Version - 6.2
• JetPack Version (valid for Jetson only)
• TensorRT Version : 8.5
• NVIDIA GPU Driver Version (valid for GPU only) :2.5
• Issue Type( questions, new requirements, bugs) bug

I can use pgie as a Peoplenet model, but I’m having trouble adding sgie as a custom model on a person label. I added config and engine files to composer, but it is still functioning normally and not producing any sgie output. just the PeopleNet model with bounding boxes is displayed.

3

What is your SGIE? What do you want to display with your SGIE’s output?

4

I have an apparel identification model used as sgie that uses a person label from peoplenet to recognise different types of clothing.
You may prefer my blog related sgie here-

5

Can you provide your gst-nvinfer configuration files?

6

yes, Here it is.
config_infer_apparel.txt (3.5 KB)

7

What are the output layers’ dimensions of your model?

8

you mean number of classes or categories to predict ?

9

No. The layers’ names and dimensions.

10

When we created app we provided dimensions like this
" -d 3,256,256 "

11

Is this the output layer’s dimension?

12

yes this is mentioned while training model

13

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

Is the output softmax values? If not, you need to customized the classifier’s “parse-classifier-func-name” function.

15

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