Skip to content

Roadmap

Strategic priorities for the herbarium OCR to Darwin Core toolkit.

Current development focus: See GitHub Projects below for detailed progress tracking across the complete herbarium digitization ecosystem.

Completed Research Contributions

  • ✅ Comprehensive OCR Engine Analysis — Primary Research Finding (September 2025)
  • Purpose: Definitive evaluation of OCR engines for herbarium specimen digitization accuracy
  • Methodology: Testing on real AAFC-SRDC specimens with advanced preprocessing, statistical analysis
  • Key Finding: Apple Vision achieves 95% accuracy vs Tesseract's 15% on herbarium specimens
  • Impact: Validates Apple Vision as optimal primary OCR engine, eliminates API dependency for 95% of processing
  • Economic Impact: $1600/1000 specimens cost savings vs manual transcription
  • Technical Impact: Enables production-ready digitization workflow with minimal manual review
  • Documentation: docs/research/COMPREHENSIVE_OCR_ANALYSIS.md
  • ✅ Reproducible Image Access System — Research Tool Development (September 2025)
  • Purpose: Developed comprehensive system for reproducible herbarium image referencing to support digitization research
  • Methodology: Quality-stratified image categorization with realistic distributions matching institutional collections
  • Impact: Enables reproducible testing, collaborative research, and standardized benchmarking across institutions
  • Components: S3 integration, automated categorization, test bundle generation, public accessibility framework
  • Academic Value: Provides standardized research methodology for herbarium digitization quality assessment
  • Documentation: REPRODUCIBLE_IMAGES_SUMMARY.md

Immediate Priorities - Stakeholder Focused

Context: System ready for production deployment, stakeholders need tangible results demonstration.

Phase 1: MVP Dataset & Stakeholder Demonstration (Week 1)

  • ✅ MVP demonstration script ready - Process 50-100 specimens for stakeholder review
  • ✅ Stakeholder progress report - For Dr. Chrystel Olivier and Dr. Julia Leeson
  • ✅ Production system validated - 95% accuracy on real specimens
  • Ready for deployment - 2,800 specimens processable immediately

Phase 2: Full Production Deployment (Weeks 2-3)

  • Process 2,800 captured photos using validated Apple Vision pipeline
  • Quality control review with Dr. Julia Leeson (Herbarium Manager)
  • Darwin Core data delivery - GBIF-ready institutional dataset
  • Complete processing documentation with audit trail

Phase 3: Institutional Integration (Weeks 4-6)

  • Database integration - Transfer to institutional collection systems
  • Staff training completion - Handover to successor workflows
  • Long-term sustainability - Ongoing digitization procedures
  • Success metrics validation - Final project evaluation

See HANDOVER_PRIORITIES.md for detailed 8-week plan.

Long-term Development Features

  • Integrate multilingual OCR models for non-English labels — Future priority (#138)
  • Integrate GBIF taxonomy and locality verification into QC pipeline — Future priority (#139)

Issue Management

Create GitHub issues from roadmap entries:

python scripts/create_roadmap_issues.py --repo <owner>/<repo> \
    --project-owner <owner> --project-number <n>

This script keeps the roadmap synchronized with GitHub Projects for automated agent workflows.

Medium Priority Features

  • Support GPU-accelerated inference for Tesseract — Q3 2025 (#186)
  • Populate mapping rules in config/rules/dwc_rules.toml and config/rules/vocab.toml (#157)
  • Audit trail for import steps with explicit user sign-off (#193)
  • Add evaluation harness for GPT prompt template coverage (#195)

For a complete feature history, see CHANGELOG.md.

Project Organization

The AAFC herbarium digitization project spans multiple domains requiring coordinated development across several GitHub Projects:

🏗️ AAFC Herbarium Infrastructure

Focus: Deployment, operations, and production workflows - Import audit workflows and compliance - Configuration management and deployment automation - Production monitoring and system integration - Multi-repository orchestration and CI/CD pipelines

💻 AAFC Herbarium Core Development

Focus: Core toolkit features and technical enhancements - OCR engine improvements (GPU acceleration, multilingual support) - Schema parsing and mapping automation - Development tooling and testing infrastructure - Performance optimization and technical debt

📊 AAFC Herbarium Data & Research

Focus: Data quality, analysis, and research workflows - GBIF integration and taxonomic validation - Geographic data verification and gazetteer services - Export formats and reporting tools - Research collaboration and data publication

📋 Legacy Project

Status: Being reorganized into the new structure above

This multi-project structure supports the full scope of herbarium digitization beyond just code development, enabling coordinated progress across infrastructure deployment, research workflows, and institutional integration.

[AAFC]: Agriculture and Agri-Food Canada [GBIF]: Global Biodiversity Information Facility [DwC]: Darwin Core [OCR]: Optical Character Recognition [API]: Application Programming Interface [CSV]: Comma-Separated Values [IPT]: Integrated Publishing Toolkit [TDWG]: Taxonomic Databases Working Group