We're setting up a test script in python on a raspberry pi fitted with an NCS2. The script successfully connects to the NCS2 via inference engine and runs the necessary inferences to detect faces using Intel's open-source "face_detection_retail" model. This is the same model used by the interactive face detection demo in the OpenVINO app zoo.
What we're trying to do now is track unique faces i.e. without necessarily identifying them we want to know that a certain face is the same face between multiple frames.
We've currently tested an IOU tracker and are getting mixed results. I was wondering if anyone here knows of any other trackers that might work well with the NCS2 and this specific face detection model?
Thank you !
Thanks Shubha. We will definitely give these a shot! Was also wondering if there is any way, per se, to run tracking computations directly on the NCS2 using the openvino SDK.. I know this forum is for openvino not NCS2 so apologies if it's off topic !