Tasks.md

Lead Eagle - Tasks

Week 1 (18-11-26)

Basic proof of concept.

• Image import
• Image segmentation
• Measurements
• Data export

Week 2 (until 18-12-04)

• Extend number of extracted features
  • Use OpenCV shape features
• Fix export format
• Learn Vue.js

Week 3 (18-12-10)

Create a basic web application.

• Fix export format (i.e. [t], [f])

Week 4 (18-12-17)

• UI: Browse & Create datasets
• UI: Upload Files
  • https://serversideup.net/uploading-files-vuejs-axios/
• Backend: Index files
• DB-Model: Dataset (ID, name, path), File (Dataset ID, subpath, modification time, size)
  • http://flask-sqlalchemy.pocoo.org/2.3/models/
• UI: Browse files
• API:
  • /api/datasets/<id>/files/<path>
  • /api/projects/<id>

Week 5 (19-01-07)

Fixes.

• Fix and cleanup Upload: Do not compress
• Fix EcoTaxa zip (only one ecotaxa_* file)
• Apply Martin's remarks
• Refactor /api into its own Blueprint to keep the main application file clean
• UI: Browse & Create projects
• Create sensible visual structure: Headline for navigation
• Remove absolute references to http://localhost:5000
• Implement Auth analogously to MorphoCluster
  • • Command for creating users

Week 6 (19-01-14)

• Pfade anpassen (Konfigurierbares Datenverzeichnis)

Week 7 (19-01-21)

• Make download work
• Include contour images in ecotaxa export
• Draw contours only for one object in single-object image ("Vignette")

Week 8 (19-01-28)

• Intra-Node computation with processing.Pipeline (see experiments.graph)
• Tests

Week 9 (19-02-11 - 19-02-22)

Week 19 (19-02-25 - 19-03-01)

• Configuration of Pipeline Nodes
  • Input / Output slots
  • Parameters
• Graylevel features
• img_rank in ecotaxa output.
• Set background to white (alpha configurable)
• Run processing in background: Specification & Implementation
  • See MorphoCluster, Readme
• Account management: Specification & Implementation
  • MorphoCluster does not have this.

Later

Implement processing nodes.

• UI: Browse & Create processing nodes
  • Api:
    • /projects/<id>/nodes/<id>
  • DB-Model: Node (ID, name, class), Edge (from ID, from slot, to ID, to slot)
  • Node 1: Dataset Source
    • Read file index of dataset
    • Parse file names
    • Extract object ID (default: filename)
    • [Extract parent object ID (for later stitching)]
    • [Use index file to provide metadata]
  • Node 2: Processing
    • Do everything (segmentation, object extraction, ...) in one step and write results to the database
  • Node 3: Export Sink
    • Generate EcoTaxa Export