I want to run primary/secondary classification on entire frame. All the samples I could find on deepstream_python_apps perform classification as secondary inference on ROIs of frame given by primary detector.
Please, clarify how classification on entire frame can be performed as primary and secondary classifier.
Can you describe your whole pipeline and the purpose? Normally you need a detector and then execute classification on the output of the detector, I don’t understand how you can execute the classification without knowing the detected object first, thanks to clarify.
My purpose includes whether, terrain type, etc classification. For such purposes I needs to perform classification inference on entire frame.
I am currently trying to modify deepstream-app2 for my purpose.