Export detections (bounding-box coordinates, classfication...) for detection/tracking

[careyer]

• rpi-deep-pantilt version: 1.2.1
• Python version: 3.7.3
• TensorFlow version: 2.4.0
• Operating System: RaspiOS lite (latest version)

Description

Searching for a way on how to export the bounding box coordintes for the detection / tracking. I want to evaluate where in the video objects are detected. Is there a way to export that information in order to process it somewhere else (e.g. by piping the output of rpi-deep-pantilt to anyother program/process?
It would be awesome if this could work with and without edge-tpu. (P.S: I noticed that the console output with edge-tpu is much less informative than without).

Other than that I'd like to trigger some action if a specific object is detected with a probability threshold >xx% , e.g. if a bird is detected trigger a Deterrent system.

Thank you very much fo the great project!