Skip to content

Latest commit

 

History

History
275 lines (199 loc) · 7.87 KB

sdr-tools.md

File metadata and controls

275 lines (199 loc) · 7.87 KB
Raw

SDR Tools Technical Reference

This documents serves to outline the tools currently included in the SDR, their purpose and interfaces.

Contents:

Teklia Machine Learning tools

Teklia DLA

Purpose

The Teklia segmentation tool implements allows the identification of regions of interest, identifying alternatively samples or lines of text.

Input

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]
  • Training model
    • Description: Selection of models for use in the Teklia tools
    • Type: [text] (options: "pinned_insects", "line_synthesys")

Output

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]

Teklia HTR

Purpose

The Teklia Handwritten Text Recognition tool reads handwritten text in identified regions utilising the Kaldi library.

Input

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]
  • Training model
    • Description: Selection of models for use in the Teklia tools
    • Type: [text] (options: "kaldi_synthesys")

Output

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]

Teklia NER

Purpose

The Teklia Named Entity Recognition tool implements the Spacy library.

Input

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]
  • Training model
    • Description: Selection of models for use in the Teklia tools
    • Type: [text] (options: "pinned_insects")

Output

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]

Third party image processing tools

Mothra

Purpose

Mothra, is a tool for automatically measuring the wingspan of lepidopterans. When provided with an image of a lepidopteran alongside scale bar, it attempts to produce a set of measurements.

Input

  • Image
    • Description: The input image on which Mothra should be run
    • Type: [PNG, JPG, JPEG, TIFF, TIF]
  • Detailed plot
    • Description: Flag to determine if Mothra should produce debugging information on output images
    • Type: [bool]

Output

  • Mothra Image
    • Description: The processed image output by Mothra
    • Type: [JPG] (collection)
  • Mothra Report
    • Description: A csv file produced by Mothra detailing the measurements that it made
    • Type: [csv]

Barcode reader

Purpose

This tool reads barcodes from specimen images using the bardecocde library.

Input

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]
  • File location
    • Description: Name of the input file
    • Type: [json]

Output

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]

Input file processing

Split File

Purpose

Splits a file line by line into seperate entries in a collection

Input

  • Input
    • Description: The file which is to be split
    • Type: [generic text]

Output

  • Split files
    • Description: The multiple output files generated by the split
    • Type: [generic text] (collection)

Create SDO from CSV (CSV to OpenDS)

Purpose

Create a specimen data object for de novo digitisation from a SDR CSV file.

Input

  • Input
    • Description: The SDR specified CSV file to be used in the digitisation
    • Type: [csv]

Output

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]

Create OpenDS

WARNING: DEPRECATED

Purpose

Create a specimen data object for de novo digitisation

Input

  • Input
    • Description: Input image
    • Type: [png]

Output

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]

Output file processing

Save output as ROCrate

Purpose

Saves the output of a digitisation to disk as an RO Crate

Input

  • Bundle Type
    • Description: Selector to specify if wishing to bundle individual datasets or a collection
      • Type: [select] (options: "Individual dataset", "List Collection")
  • if Individual dataset, Single Files
    • Description: Specify one or more files for bundling
    • Type: [data]
  • if List Collection, Data Collection
    • Description: Specify a collection for bundling
    • Type: [data] (collection)

Output

  • ROCrate
    • Description: A Research Object Crate containing the input datasets / collections
    • Type: [zip]

OpenDS to CSV

Purpose

Create a flat csv from a specimen data object

Input

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]

Output

  • Input
    • Description: A flattened representation of the OpenDS object
    • Type: [csv]

Save Output

Purpose

Saves the output of a digitisation to (the server's) disk

Input

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]
  • Batch ID
    • Description: A user supplied identifier, under which to save the output
    • Type: [text]

Output

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]

Misc

Validate OpenDS

WARNING: DEPRECATED

Purpose

A tool to validate a specimen data object.

Input

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]

Output

  • Validation report
    • Description: Success or failure of validation
    • Type: [text]

Download Image###

WARNING: DEPRECATED

Purpose

Download an image to the local filesystem and append file info to opends object.

Input

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]

Output

  • OpenDS specimen object
    • Description: Incrementally built Fair Digital Object collating specimen data and annotations
    • Type: [json]