How to extract images from stream to feed into sgie

[damonzzz]

Please provide complete information as applicable to your setup.

**• Hardware Platform GPU
**• DeepStream Version 6.0
**• TensorRT Version 8.0.1
**• NVIDIA GPU Driver Version 11.4
**• Issue Type questions

Hi there.

I’m trying to use yolov5 and retina face to extract facial features. I tried build a pipeline with yolo as pgie and rface as sgie, it worked fine.

However, considering for some memory and computing performance, I need to extract some certain images/frames (e.g an obj with a new track_id) from the rtsp stream and feed them into sgie.
Is it possible? If it is, plz give me some hints.

[kesong]

You can add one probe function and filter out some object meta data.

[damonzzz]

thanks for reply. I solved it by using probe function as you suggest.

Now I have a new question, could you please have a look and give me some hints?

https://forums.developer.nvidia.com/t/how-to-extract-facial-features-using-arcface-tensorrt/204017?u=damonzzz