What it does
ilastik is an interactive learning and segmentation toolkit for classifying pixels or objects, tracking and counting experimental image data.
In Pixel Classification, choose features and brush-label representative pixels, then correct training using live predictions. This produces semantic classes and probabilities, rather than separate object identities. Prediction Export saves multichannel probability maps or class-label images, with format, data type, range, axis order and destination controls. Viewer screenshots are separate outputs. Apply a trained classifier to new images in batch/headless mode, and retain project/training choices with the data. Delete All in the export applet removes generated output files.
Strengths
- Interactive brush-based training
- Pixel and object classification workflows
Best for
- Building a research-image segmentation workflow
Sources
- ilastik - ilastik (Oct 26, 2026)
- ilastik pixel semantics and batch output (Oct 26, 2026)
- ilastik data export versus viewer capture (Oct 26, 2026)