How to Retrieve the Parent of object_meta When Using Parallel Inference in DeepStream?

[lqdisme]

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) GPU
• DeepStream Version 6.3
• NVIDIA GPU Driver Version (valid for GPU only) 530.30.02
• Issue Type( questions, new requirements, bugs) questions

I am currently working with a pipeline in NVIDIA DeepStream SDK that includes parallel inference. One branch of my pipeline consists of the following components: PGIE → Tracker → SGIE.

My issue arises when I try to retrieve the parent of an object_meta while using a probe function on the metamux source pad. Specifically, I need to determine which parent (object) an object_meta is associated with after it passes through the tracker and SGIE in parallel inference.

Could someone guide me on how to efficiently retrieve the parent object_meta in this setup? Are there any specific properties or metadata I should focus on when using the metamux source pad?

Any advice or example code snippets would be greatly appreciated. Thank you!

[yuweiw]

Could you attach your whole pipeline? The parent object_meta may be cleared after some plugins, such as nvvideoconvert.

[lqdisme]

My pipeline is similar to the example of a parallel inference pipeline. It has two branches: Branch 0 consists of PGIE0 → Tracker0 → SGIE0 (object detector), and Branch 1 consists of PGIE1 → Tracker1. When I retrieve the metadata from the metamux source pad, I cannot access the parent object_meta. It seems that metamux does not retain the parent information, similar to nvvideoconvert. Could you please give me some advice?

[yuweiw]

Like nvvideoconvert plugin, we cannot support get the parent after the metamux plugin currently.