Hi,
I’m finally getting around to updating some old workflows and it does seem like the Sobel score works better than other methods for auto-selecting 2D classes, so I was hoping to update our automated “default” processing workflows. However, it is not ideal for an automated workflow that the user has to manually input a circular mask diameter or the job will not run. I am wondering if a future iteration of this job could make a default mask size similar to the window size in 2D classification (85% of the box size or something similar) and then a user can manually change it if desired – this would make implementation of the method in automated workflows much easier / more versatile. We don’t have masking at the 2D classification step in our default workflows, so the sharp edges situation doesn’t apply and I imagine probably wouldn’t in many automated workflows.
Thanks!
Talya