Deepstream-app objectDetector_Yolo add secondary inference

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Hi

I am using objectDetector_Yolo with two custom model.

I successfully changed config file to use one model.

I want to know how to add one more model for inference.

I see that deepstream_app_config_yoloV3.txt has [primary-gie]. What should be added to use one more model?

Thank you.

3

I checked https://docs.nvidia.com/metropolis/deepstream/4.0/Custom_YOLO_Model_in_the_DeepStream_YOLO_App.pdf and in step 3 I need to change the kANCHORS.

If the kANCHORS is different for the two models, how should I set it?

4

Hi @pijebom737,
Sorry for delay! the 2nd model is not YoloV3, right?
You could refer to deepstream-test2, replace the first model with YoloV3 and the sgie1 model to your model, and remove other sgie models.

If the kANCHORS is different for the two models

Do you mean the 2nd model also use kANCHORS ?

1 Like
5

Hi @mchi
How to do

replace the first model with YoloV3 and the sgie1 model to your model, and remove other sgie models.

7

@PhongTN ,
Refer to the YoloV3 config file and the code the deepstream-test2.

Thanks!