Hello,
Is there a method where I can perform my own manipulation of the streams frame by frame such that the algorithm modules (specifically Hand Tracking) would be aware of and work with the custom segmented image?
Better yet, is there a method built in to the SDK to allow me to supply a reference image to train the SDK to ignore.
More info in my needs. I need high accuracy hand recognition and tracking in a fixed frame/fixed background environment. However this setup includes situations where the SDK struggles to correctly recognize the hand or recognizes background objects as hands.
