HTRflow 🔍
HTRflow Demo
Developed by the National Archives of Sweden with Huminfra that demonstrates AI-powered conversion of historical manuscripts to digital text using HTRflow.
Note: This demo application is for demonstration purposes only and is not intended for production use. The application is hosted on Hugging Face 🤗 using shared infrastructure, which means there is a daily quota limit on how much you can use the app each day.
Both the App and HTRflow's code are completely open source. Explore and contribute on GitHub:
If you find our projects useful, please consider giving us a star ⭐!
Contact
Upload
Select or upload the image you want to transcribe. Most common image formats are supported and you can upload max 5 images at a time in this hosted demo.
Settings
Select a pipeline that best matches your image. The pipeline determines the processing workflow optimized for different handwritten text recognition tasks. If you select an example image, a suitable pipeline will be preselected automatically. However, you can edit the pipeline if you need to customize it further. Choosing the right pipeline significantly improves transcription quality.
Result
The image to the left shows where HTRflow found text in the image. The transcription can be seen to the right.
Export
Choose file format for export.