hare

I'm trying to somehow label/"recognize" faces from the cv.jit.faces object.

The installation I'm working on will automatically crop out recognized faces and save them as a still graphic - but the problem is not to save the same face over and over in consecutive frames.

The only solution i can think of is to just use the draw function and attach a tracking point with cv.jit.track to the "face rectangle" the moment the frame is captured as a still. (Then compare coordinates of tracking points with face coordinates before saving the frame)

But that's actually not very elegant and would also result in pretty high load for my poor macbook which will be running some other applications on the side…

The only solution i can think of is to just use the draw function and attach a tracking point with cv.jit.track to the "face rectangle" the moment the frame is captured as a still. (Then compare coordinates of tracking points with face coordinates before saving the frame)
But that's actually not very elegant and would also result in pretty high load for my poor macbook which will be running some other applications on the side…

cv-jit-faces-labeling-the-faces

i once used a combination with the blobs.sort, i couldn't find the patch anymore but the principle worked.

cv.jit.faces - labeling the faces?