-
Notifications
You must be signed in to change notification settings - Fork 132
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
pop index out of range error in pixelclassification workflow #1244
Comments
I can confirm the issue still persists in 1.3.2 rc1. Encountered this exact issue in the autocontext workflow when trying to save predictions as zyx instead of zyxc. |
thx a lot for checking @snajder-r |
can confirm this still is an issue related (in a way): ilastik/volumina#263 |
I just ran into this and this when trying to export a prediction (Simple segmentation). The original image is Once I get the "pop index out of range IndexError" error, there seem to be no way back - weird errors start to happen. Even and attempt to delete all exports (the "Delete All" button) causes an error:
Moreover, (it looks like) I was stupid enough to save the project after the error. Thankfully, the fix was pretty easy. Just in case anyone would need this: import h5py
input = r'C:\project\project.ilp'
with h5py.File(input, "a") as f:
del f['Prediction Export'] The versions of the python packages I used:
Based on "C:\Program Files\ilastik-1.4.0.post1-gpu\conda-meta"). Ilastik 1.4.0.post1. |
There is an issue with dropping the channel axis during prediction export in the pixelclassification applet.
How to replicate:
During prediction export if the axis order in the image export options is set from e.g. yxc to yx (dropping the channel) after closing the options menu saving the project leads to an "pop index out of range IndexError". If the project is closed in that condition it can't be opened again.
The only work around to save the labels was to manually add an empty channel axis to the hdf5 project file. After that the project can be opened again.
Image of the error:
Cheers, Adrian
The text was updated successfully, but these errors were encountered: