Remain tracking_id object_id in multi detection

**• Hardware Platform (Jetson / GPU): Jetson Nan
**• DeepStream Version: 6.0.1
**• JetPack Version: 4.6
• TensorRT Version 8.0
**• Issue Type: Bug
**• Iam using deepstream to make face recognition app.
The pipeline I was using is:

pgie (person detect) → tracker → sgie1 (face detect) ->sgie2 (extract face embedded)

However I am facing a problem, the TD (tracking_id from object_id) pgie (person detect) and sgie1 (face detect) were different.

image

I also used this parameters in config file for sgie1 but it didn’t work
image

How can I remain the object_id from pgie to sgie1 in this case ?

From you pipeline, the tracker only tracked the person. So the object_id of face object should be invalid.

Suppose the face object has parent object (person object). Seems you can find the object from parent object.

I’m understanding that the results of pgie and sgie are being included in the same obj_meta_list, so how do I know who this face’s recognition results belong to?

I realized that obj_meta.parent can be used to get the object_id in first pgie, however when calling pad in osd, obj_meta.parent always returns None

This is my pipeline

Here is where I log out the result, always return None

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks

Can you have a try probe the gstbuffer at the src pad of the SGIE?

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.