ilastik

ilastik is a desktop tool for researchers who need to identify, classify or count cells and other objects in experimental images. It runs on Linux, macOS and Windows. Its main draw is interactive training: you mark examples in an image with brush strokes and see predictions as you work, without needing machine learning expertise.

Pixel classification assigns image regions to classes and uses predicted class probabilities for semantic segmentation. You can choose predefined image features rather than design them yourself. Object classification predicts the classes of whole objects, while the carving workflow helps extract individual objects from a 3D volume with some manual guidance.

For cell analysis, ilastik can count densely packed cells and track dividing cells in both 2D and 3D data. Tracking supports automatic and manual work, so researchers can use either approach within the same toolkit. Most operations remain interactive even with large datasets, which matters when you're assessing predictions and refining labels against the images themselves.

The emphasis is on visual feedback and image analysis tasks rather than requiring researchers to write a machine learning pipeline. An extensive user manual covers the workflows, and the image.sc forum provides a place to discuss questions with other users.