Machine learning image to category

Hello. I have been using the Weka segmentation. So far it has helped me in auto thresholding the images. I was wondering if there was a way to assign a whole image to a category by machine learning. I have images of different samples, and each of these images (samples) has a structure that can potentially be correlated to another parameter (ie. high, low, med-high…). Thank you!

Hi @I.G,

I’m afraid there is no image classification plugin in ImageJ yet. If you can segment objects in the image with machine learning and want to classify the object, you may want to give #ilastik a try. It supports object classification:
https://www.ilastik.org/documentation/objects/objects.html

But out of curiousity: Can the structure be quantified (size, shape, …)? Maybe you can classify the objects according to such measurements?

Let us know if this helps.

Cheers,
Robert