Deepstream 5.0 operate classifier on another classifier's results

Hello!
I have a pipeline “detector → classifier_1 → classifier_2”, and i want to operate second classifier on a certain class id from first.
For first classifier gie-unique-id = 2, and for second i set operate-on-gie-id = 2, operate-on-class-ids = 0, but got no results from classifier_2. Classifier_1 works fine. Tried to remove operate-on-class-ids, still no results. Are there any possibility to do so? Or i have to operate both classifiers on detector and then change meta by myself?
System:
Jetson AGX Xavier
Python 3.6
Jetpack 4.4.1
TRT 7.1.3
Deepstream 5.0.1

can you upload your config

classifier_1.txt (494 Bytes) classifier_2.txt (498 Bytes) detector.txt (472 Bytes)

Hey customer,
Sorry for the delay, can you explain more about why your 2nd classifier need to perform on the 1st classifier?

I need to perform second classifier only on bboxes, with certain label from first classifier, not on all bboxes. And in my case its better, than making more different classes in detector.

Hey customer, sorry for the late.
Just to confirm, I checked you config files and saw both sgie1 and sgie2 are performing on the class-id=0 from pgie,
So sgie1 will do some classifier on class-id=0, let’s say A, B, C, so you just want to make sgie2 only perform on the class-id=0 from pgie && sgie1 classifier result, assume A, right?

Yes, u are right. Or maybe i can do it with another way, change object class_id after sgie1 to newly created, that is not in detector’s classes list, in probe, placed after first classifier, but i dont think it’s optimal solution.

It’s a good idea to do it for the current DS SDK.

We will consider if we need to add the feature in later DS release.

Hello,
we have a similar problem, we need to apply second classifier to only one class of first classifier. what is the best practise to do this?

@behna.rahimi you can refer this idea, please create a new topic if you have issues.