I recently purchased a luxonis camera for obstacle detection. I was thinking that I could possibly modify the YOLOv8 architecture so that it can take in a depth map as the fourth channel in addition to the 3 RGB channels (mostly to deal with objects that are similarly colored to the background). However, I haven't been able to find much mention of this online. Has anyone ever heard of something like this being done before? Is it possible, and would it be worth the effort?

    drago627
    I believe it is possible, but you'd need to use a NeuralNetwork node, instead of Yolo node. Also, you will need custom decoding.
    Would it be worth the effort? - sure if you already have the model and it has shown better accuracy than the standard RGB models.

    Thanks,
    Jaka