Question about Secondary GIE

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) Jetson Xavier NX
• DeepStream Version 5.1
• JetPack Version (valid for Jetson only) 4.5.1
• TensorRT Version 7.1.3
• NVIDIA GPU Driver Version (valid for GPU only)
• Issue Type( questions, new requirements, bugs) Questions

I have some questions about how secondary GIE works. In my project I have a primary GIE that detects people. Then I have another model for mask detection as secondary GIE. In configuration file for secondary GIE model, I have operate on class for “person” class. But I have some questions how this pipeline works:

  1. What is the input for secondary gie? The entire frame with a filter applied for the chosen class (“person”) or the cropped image (the bounding box) of detected people?

  2. Can I choose if i want secondary gie to detect on the entire frame or the person’s bounding box? If so, what is the configuration field for that?

  3. Is pyds.NvDsFrameMeta.source_frame_height the frame height that is the height of the frame given as input for the object detection model?

Thank you!

Cropped ROI

check process-mode and operate-on-class-ids , refer Gst-nvinfer — DeepStream 5.1 Release documentation

No, generally, it’s the frame width/height set in nvstreammux

1 Like