How to apply a mask to primary detections before feeding into secondary detector

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Please provide complete information as applicable to your setup.

• Hardware Platform (GPU)
**• DeepStream Version 7.1
• JetPack Version (valid for Jetson only)
• TensorRT Version 10.3.26
**• NVIDIA GPU Driver Version (valid for GPU only) 560.35
• Issue Type( questions, new requirements, bugs)

Hi, I am trying to build up a deepstream pipeline. I am having input frames through camera input and then I am feeding it to my primary detector to detect the objects that I want to perform segmantation task. And then I am feeding those objects into my secondary model for segmentation. The only problem is I want to set pixels that does not contain my object to ‘0’ so second model gets the desired object with black background to perform the segmentation. But I couldnt find a way to apply this mask on primary detections. Can you guide me?

Thanks in advance.

3

We cannot support this scenario at present. You can only use segmetation without the primary detector to implement your needs.