When the WSI was scanned by factor of 40x, the cell looks bigger than it is original looks. Just like the pictures attached. In the after.jpeg, qupath detect cells wrong. so how can I use QuPath to detect cells correctly?
So when you are detecting cells there are a lot of parameters to change that help for account for the differences between types of images. This will especially be true if your images aren’t scaled and you are going by pixels due to images not having scaling. You should have a look in the documentation about this but one obvious example are the min/max filters on size
I’ve tried but it’s quite tough on the JPEG you uploaded in the time that I had, hopefully you can get a better result on your real data.