LARC-CMU-SMU/FoodSeg103-Benchmark-v1

dataset about FoodSeg103

Opened this issue · 1 comments

Dear scholars,
I am very excited and thankful for you job, the new food dataset you have proposed is very important for the food image segmentation, but I have a questions about this dataset, I found the mask(annotations) in your paper is RGB image, however, after I have downloaded this datasets, I found the mask(annotations) is single channel. So I want to ask how you made it?

@Mark1Dong In visualization step, each ingredient class corresponds to a specific color, and these color maps are only used for visualization. It's a quite easy script and of course you can try different class-color mapping function from ours.