Processing live video stream from opencv on raspberry pi 5

Let me elaborate a little more.

I need to add the tracker analyzer

tracker = degirum_tools.ObjectTracker(
class_list=[“person”],
track_thresh=0.35,
track_buffer=100,
match_thresh=0.9999,
trail_depth=50,
anchor_point=degirum_tools.AnchorPoint.TOP_CENTER,
)

This is how my code was working before

for detected_persons in degirum_tools.predict_stream(person_detection_model, video_source, analyzers=[tracker]):

Now I want to integrate camera rotation so I followed the example and I realize that the method predict_stream is not used in the given example

its using predict_batch instead, If I try to pass the analyzer as follows:

for results in person_detection_model.predict_batch(rotated_frame_generator(video_source), analyzers=[tracker]):

it throws an error

TypeError: Model.predict_batch() got an unexpected keyword argument ‘analyzers’

The question is How to integrate analyzer with predict_batch?

@shashi

I just realize how to approach it by

degirum_tools.attach_analyzers(model, [tracker])

Thanks

1 Like