Kraken review
A trainable OCR system for specialists working with archives, handwriting, and non-Latin scripts.
Reviewed by iTechGuides Editors · Editorial team · Updated Oct 2026
Kraken is an open-source OCR and automatic text recognition system for historical documents, handwriting, and non-Latin scripts. It is suited to digital-humanities specialists, archivists, researchers, and developers who need to adapt recognition and layout models to particular materials. Kraken can process images and PDFs locally on Linux and macOS, with command-line and Python API interfaces for building repeatable workflows.
Its main strength is trainability. The software supports trainable layout analysis, reading-order detection, segmentation, and character recognition, including model training and fine-tuning for specific scripts or handwriting. Support for right-to-left, bidirectional, and top-to-bottom writing makes it suitable for documents that require more than standard left-to-right recognition. Output options include ALTO, PageXML, AbbyyXML, and hOCR, while word bounding boxes and character cuts provide structured results for archival and research workflows.
Kraken also fits established humanities-oriented ecosystems through integrations with eScriptorium, OCR4all, OCR-D, and Arkindex. Command-line workflows can be chained, and the Python API covers preprocessing, segmentation, recognition, and serialization. The trade-off is a specialist orientation: users looking for a simple, general-purpose desktop OCR workflow may find the command-line and programming interfaces less approachable. Kraken is a strong choice when trainable models, historical materials, complex writing directions, or archival output formats matter; a more general end-user OCR tool may be a better fit for routine document conversion.
Kraken pros and cons
- Where it wins
- Trainable layout, reading-order, segmentation, and character models
- Supports handwriting and right-to-left, bidirectional, and vertical scripts
- Exports ALTO, PageXML, AbbyyXML, and hOCR with detailed recognition data
- Where it doesn't
- Requires technical workflows through command-line or Python interfaces
- Focused on specialist OCR rather than general end-user document workflows
- Runs locally on Linux and macOS rather than across a broad platform range
Kraken fact sheet, pricing and score →
Advertiser disclosure: iTechGuides is reader-supported. Vendors can pay for top positions in our rankings and for a place on other products' pages, and we may earn a commission when you click some links. Bought places stand at the number bought; this page and our disclosure say so. Paying vendors may get a fuller, more detailed review from our editors, and that review can change their score. How we rank.
Last updated · How we research and update