Import caffe model in deepstream problem

• Hardware Platform (GPU)
*• DeepStream Version: 5.0-20.07-devel
• TensorRT Version: TensorRT 7.0
• NVIDIA GPU Driver Version: Driver Version: 440.100 CUDA Version: 10.2

Hello everyone, I am trying to import two secondary classifiers (gender and age), to be run on a face detection model. The face detector model is one provided by nvidia and it works perfectly, however my classifiers don’t work, deepstream generates the engines correctly, but at the time of inference it only detects class 0, or nothing.

I use the following pipeline

Pgie (Face detector) -> Tracker -> Sgie1 (Age), Sgie2(gender)

Here I have all the files I need for the deploy, both the models and the deepstream configuration and test video.
Repository test

If you could give me some guidance as to what I’m doing wrong, I’d be grateful.


Hi @claudio.campuzano.b
Could you test again without tracker?


I’m trying to use your models too, but I can’t see age and gender, only the faces …
Tracker is disabled.
How can i do?
i am using deepstream-app -c …

Hi @andrea_vighi,
This ticket is closed, if you want to chase this issue, please file a new ticket with detailed repo step.


1 Like