Hey guys,
Before I go forking and modifying the code to add this feature -
is there any option to add a fourth input, the depth value, per each pixel/object in training a detection YOLO?
Has it ever been implemented before?
Example of the depth values of an image:
Thanks 
Ultralytics only supports 3 channels by default. So you would have to modify it (probably to a significant degree especially when it comes to augmentations) to get it to work with 4 channels.
Thank you.
I’m willing to give it a try, as I’m enthusiastic about learning about the framework and the under-the-hood technology through action.
If you’re down to guide me a bit on what code segments of the pipeline should be modified, I’ll be more than happy to ask from time to time about the steps I should go through to ensure the modifications are complete.
In the meantime, I’ll try mostly to chat with GPT to get a hold of what’s going on there.
