A scientific tool formHTR (https://github.com/grp-bork/formHTR/) was developed for extraction of handwritten text from paper forms (logsheets) of known structure. The tool uses advanced pre-trained OCR models for text recognition and extraction. To subsequently organize the extracted data, it is necessary to annotate predefined PDF templates with expected features and their positions. The scanned logsheet is then aligned with the template to aid the extraction of data from the defined positions. The results are typically stored in a tabular form to allow manual proofreading afterwards.

The tool currently only provides a command line interface, which is hardly optimal for the target audience of biologists. The goal is to develop an interactive web-based interface that will:

• Create/update logsheet document templates.
• Annotate PDF templates with positions of known elements using interactive editor.
• Execute content extraction functionality and manual alignment of documents based on annotated element position.
• Provide proofreading mechanism allowing to double check all extracted values and correct them if needed.

The topic has an external consultant (formHTR autor) who will provide necessary data and technical specification for this task.