colorbar_params in overlay_prediction_mask #410

• TIA Toolbox version: 1.2.0 (matplotlib 3.5.2)
• Python version: 3.8.13
• Operating System: Debian WSL2 on Windows 11 through visual code studio

Description

Reproduce the colorbar of the WSI inference plot of example notebook 6 (https://github.com/TissueImageAnalytics/tiatoolbox/blob/c773e1cd4d30d7b22e5d192fa2784b172d2b68a8/examples/06-semantic-segmentation.ipynb)

The color bar showed only 2 colors, mapped to the wrong class. I expected the same 5 colors mapped to the right class as seen in the example.

I am not sure if this is related to versions, my setup or that this is also happening to other people. My solution seems to do no harm by specifying explicitly the values. But I wanted to check that this is real or only happening to me before adding a fix.

What I Did

According to https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.colorbar.html the values property could (should?) be set with indexed colors.

In function overlay_prediction_mask, I added "values": uid_list

colorbar_params = {
"mappable": mpl.cm.ScalarMappable(cmap=cmap),
"boundaries": uid_list + [uid_list[-1] + 1],
"values": uid_list,
"ticks": [b + 0.5 for b in uid_list],
"spacing": "proportional",
"orientation": "vertical",
}