PaliGemma returns the following output for segmentation of an image:

I understand the first 4 entries are the coordinates of the bounding box, and from here I understand how to convert them into proper coordinates:

https://huggingface.co/blog/paligemma

But I have been completely unable to deal with the segmentation mask. I can find no documentation that hints even slightly at how one would interpret those numbers, I brute-forced every approach I could think of, but nothing came up with a proper mask.

Seldom have I felt so dumb in my life, certainly this is obvious and I am just missing something that literally every other sentient being on earth understands?