ilastik is a free and open-source software tool for interactive image classification and segmentation, with a particular focus on (bio)image analysis. It was first released in 2011 by scientists at the Heidelberg Collaboratory for Image Processing (HCI), University of Heidelberg, and since 2018 has been maintained by Anna Kreshuk's group at the European Molecular Biology Laboratory (EMBL).
Features
ilastik allows users to annotate an arbitrary number of classes in images using a mouse interface. Based on these user annotations and generic (nonlinear) image features, the user can train a random forest classifier. Trained classifiers can be applied to new data via batch processing or in headless mode (without the graphical user interface). Key capabilities include:
- Pixel Classification – classify each pixel in an image based on user-drawn annotations.
- Autocontext – iterative classification that refines predictions.
- Object Classification – classify segmented objects.
- Tracking – track cells or other objects over time (including animal tracking).
- Density Counting – estimate object counts from images.
- Carving – interactive segmentation of 3D volume data.
- Boundary-based Segmentation with Multicut – segmentation based on boundary predictions.
- Neural Network Classification – support for GPU-accelerated deep learning prediction.
Integrations
ilastik can be integrated into various related tools:
- Pre-trained workflows can be executed directly from ImageJ/Fiji using the ilastik-ImageJ plugin.
- Pre-trained Pixel Classification workflows can be run in Python via the ilastik Python package (available through conda).
- ilastik has a CellProfiler module for processing images within a CellProfiler framework.
Technical Details
- Stable release: 1.4.1 (May 31, 2025)
- Operating system: Cross-platform (Python-based)
- License: GPLv2
- Repository: github.com/ilastik/ilastik
Publication
ilastik was described in a 2019 Nature Methods article titled "ilastik: interactive machine learning for (bio)image analysis" (doi: 10.1038/s41592-019-0582-9), which highlights its role in bringing machine-learning-based image analysis to users without substantial computational expertise.
