What's needed for future batch-processing of new data?

Dear all,

I’m relatively new to batch-processing and I need a pointer in the right direction.

I’ve trained a pixel classifier a few months ago and would like to re-use it on new data. However, when I open the old project, my training is gone and I’m unable to load data to batch-process.
I’m not sure what files from the original training I’m supposed to load or import to be able to batch process new data. Can someone help me out or point me to the right documentation about this?

Thanks a lot!

Hi @Anjalie,

most likely ilastik is trying to load the data, the classifier was trained with. I can only assume the data, or the project have been moved. Per default ilastik saves the relative path to the data (which should also be shown in this error message that pops up when data cannot be found). So if you are unable to locate that data, ilastik GUI will not open. So you either have to go through each of the input files when starting the gui and locate them using the file browser that pops up, or restore the folder hierarchy you had in training.

There is also the option to store the data inside the ilastik project file in order to prevent such problems see here, storage. Your project file size, however, will get larger. So this isn’t possible for all kinds of data.

You can still batch process your data, but not using the ilastik gui. You would have to go via the ilastik headless mode. This involves calling ilastik from a terminal (which will be slightly different, depending on which operating system you are working in).

Just write here if you need further information on this, or if you have trouble setting up the batch processing via the headless mode!
Cheers

Dear Dominik,

Thanks a lot for the speady reply!

I thought it might be something like that. My light sheet data is indeed too large to store in the ilastik project file but I will take extra care to not move things around. Or try the headless.

Cheers!