An apology first
This post should have been written a year ago. The two workshops it describes took place in August and September 2025, and a report in Japanese appeared in the newsletter of the Japanese Society for Bioinformatics (JSBi) in Jan…

The last post took one published model and one dataset and asked where the model is weakest. This one steps back and opens the box: everything rplatypus and pyplatypus offer for semantic segmentation - a mask per class, scored by overlap - and …