Closed jluethi closed 11 months ago
(we'll need to re-evaluate this bug report. We had an issue with MD ROIs having y & x shifted. But I'm still suspicious that the thing above could be going wrong)
Never mind, it actually works as expected 🙈 (once we have the ROI bug fixed, see fractal-faim-hcs package)
The issue comes from here: https://github.com/fractal-analytics-platform/fractal-tasks-core/blob/cafe063f2a46355e529f914ea7c2f2ec9fe9cbd0/fractal_tasks_core/tasks/cellpose_segmentation.py#L435
We create the new label dataset based on the attributes of the image dataset. But MD image datasets will be yx (for the 2D case), while the output that the cellpose task generates will be zyx (also for the 2D case).
Another instance of https://github.com/fractal-analytics-platform/fractal-tasks-core/issues/150
Apparently, our most recent changes to cellpose broke the (admittedly somewhat hacky) solution we implemented in https://github.com/fractal-analytics-platform/fractal-tasks-core/pull/403
cc @fstur