I have a pallet of boards, all the same color and different sizes, stacked on top of each other. My camera is directly over the pallet looking down on it, sometimes a smaller board sits on top of a larger board and the neural net will only detect the larger board. Is it possible to use the depth camera to find the board at the highest point and then set the roi to look an inch down from that point (board height is an inch), So I would almost be performing a scan down the pallet layer by layer. Any guidance in the right direction would help.

Thanks

    Hi LeviStrieter
    Could you add a picture of what you are trying to achieve with scanning down layers. Detecting the highest point should be no problem, but I'm not sure how this would solve it. What model are you using, if Yolo, you can play around with anchors to make sure the smaller boards get identified as well.

    Thanks,
    Jaka