Semantic labels issue

[MohammadElSakka]

Hello, First of all, thank you for your work it is quite impressive. I am interested in doing image segmentation on the semantic part of the dataset, as far as I understood, the images are supposed to contain 3 classes: soil, crop, weeds. However, upon calling np.unique on some labels, I get more than 3 classes, e.g. 5.

Here is an example:

(array([0, 1, 2, 3, 4], dtype=uint16),
 './PhenoBench-v110/PhenoBench/train/semantics/05-26_00201_P0034279.png')

Could you provide me with more details?

Edit: are partial-crop (3), partial-crop (4) considered different classes or are they joined to crops (1)?