Toward a service-based workflow for automated information extraction from herbarium specimens

Area of Science:

• Botany
• Digital Humanities
• Bioinformatics

Background:

• Herbarium collections are digitizing millions of specimens globally.
• Manual extraction of label data from herbarium specimens is a significant cost driver.
• There is a need for efficient methods to capture metadata from digitized specimens.

Purpose of the Study:

• To develop and implement a formal, extensible workflow for automated metadata acquisition from herbarium specimen images.
• To integrate various automated image analysis services for streamlined data capture.
• To reduce the manual effort and cost associated with herbarium specimen data acquisition.

Main Methods:

• Developed a formal workflow integrating automated specimen image analysis services.
• Implemented the workflow using OpenRefine with a custom plugin for service integration.
• Integrated optical character recognition (OCR) for text extraction from specimen images.
• Developed methods for identifying regions of interest and extracting relevant information from OCR output.

Main Results:

• Successfully implemented an automated workflow for herbarium specimen data acquisition.
• The system performs OCR on specimen images and extracts key information.
• The workflow integrates multiple automated services for efficient metadata capture.
• The StanDAP-Herb project developed these optimizations.

Conclusions:

• The developed automated workflow significantly streamlines the capture of herbarium specimen metadata.
• Integrating image analysis and OCR reduces manual labor and costs in digitization efforts.
• This approach enhances the efficiency of data acquisition from digitized herbarium collections.