Object tracker on video

AmunTimilsina

The API documentation on Object tracker on video states in line 68 that xinFrame.out.link(objectTracker.inputTrackerFrame) is used. However, shouldn't it be manip.out.link(objectTracker.inputTrackerFrame)? This consideration arises because, as the detection_network is provided with the frame after manipulation, the tracker should ideally receive the same manipulated frame instead of the original frame from xinFrame. Am I overlooking any crucial details?

jakaskerl

Hi @AmunTimilsina
It doesn't need to be the same since the inputs will get scaled accordingly. This is so if you use a small input layer for detection (like 240x240), you are still able to view the tracking on the larger frame.

https://docs.luxonis.com/projects/api/en/latest/components/nodes/object_tracker/#reference

Thanks,
Jaka

AmunTimilsina

Thank you. Got it.