Trouble exporting batch processing data (.csv) after cell counting

Dear Ilastik team,

I’m using Ilastik to do cell density counting. The training part goes very well and when I export the .csv file I can see the different density values for my images.
However, when I try to do batch processing on a lot of images, the .csv document only shows the last value of the last image analyzed … Do you know what I can do to have the values (density/counting) for each image? Should I change something in the Export Image Settings?

Thank you for your help.

Hi @Julie,

great that you want to process many images with the counting workflow. And you ran into one of it’s limitations I am afraid - when used with the batch processing applet, the table will get overwritten. Inspired by your report, I opened an issue about this here.

What you can do, however, is use ilastik’s headless mode. In the headless mode you call ilastik from the terminal with a trained project file, and you can specify the input image, as well as an output table file. This way you would generate a new table for each image. I don’t know how familiar you are with using a terminal/command line and with scripting, but I could help get you going, if you choose to go that way.


Hi @k-dominik,

I came across this thread and am attempting to do a very similar thing. Ideally I’d want to load a series of image stacks in the headless mode and for each stack get a table of counts at composed of values for each slice within the stack.

Is this possible or will each slice be a separate .csv file that I will have to concatenate after.


Thanks @k-dominik for your help.

For info, I did my analysis on the last version of Ilastik (1.3.2) and the batch processing works with a complete .csv file (table of each image name + counting).


1 Like

Hi @a-longhini,

as @Julie pointed out, you’ll have to concatenate your csvs after. Do you need assistance with that?


Thank for the reply @k-dominik,

No, I should be able to do that part pretty easily. It’s just good to know so I don’t waste time trying to get the headless mode to do this.