In the readme.md, it says the semantic.png stores an 8-bit number indicating the nyuv2 label-ids. But actually in the dataset, the semantic.png stores a 4-channel color for each pixel. It seems that one color denotes a certain category, but could you please provide the color-category mapping list? Thank you!
In the readme.md, it says the semantic.png stores an 8-bit number indicating the nyuv2 label-ids. But actually in the dataset, the semantic.png stores a 4-channel color for each pixel. It seems that one color denotes a certain category, but could you please provide the color-category mapping list? Thank you!