ocrd-tool.json

{
  "version": "1.0.0",
  "git_url": "https://github.com/OCR-D/core",
  "dockerhub": "ocrd/core",
  "tools": {
    "ocrd-dummy": {
      "executable": "ocrd-dummy",
      "description": "Bare-bones processor creates PAGE-XML and optionally copies file from input group to output group",
      "steps": ["preprocessing/optimization"],
      "categories": ["Image preprocessing"],
      "input_file_grp_cardinality": 1,
      "output_file_grp_cardinality": 1,
      "parameters": {
        "copy_files": {
          "type": "boolean",
          "default": false,
          "description": "Whether to actually copy files (true) or just create PAGE-XML as a side effect (false)"
        }
      }
    },
    "ocrd-filter": {
      "executable": "ocrd-filter",
      "description": "Bare-bones processor can be dynamically configured to remove segments based on XPath queries",
      "steps": ["recognition/post-correction"],
      "categories": ["Quality assurance"],
      "input_file_grp_cardinality": 1,
      "output_file_grp_cardinality": 1,
      "parameters": {
        "select": {
          "type": "string",
          "default": "//*[ends-with(local-name(),'Region')]",
          "description": "Which segments to select for removal. An XPath 2.0 query expression (path and optional predicates), with 'pc' as namespace prefix for PAGE-XML and our extension functions (see help text). Only selection of segment hierarchy elements is allowed (so e.g. `*` would be equivalent to `pc:NoiseRegion|pc:LineDrawingRegion|pc:AdvertRegion|pc:ImageRegion|pc:ChartRegion|pc:MusicRegion|pc:GraphicRegion|pc:UnknownRegion|pc:CustomRegion|pc:SeparatorRegion|pc:MathsRegion|pc:TextRegion|pc:MapRegion|pc:ChemRegion|pc:TableRegion|pc:TextLine|pc:Word|pc:Glyph`, but `pc:MetadataItem` or `pc:Border` or `pc:Coords` would not match).\nFor example, to remove words or glyphs with low text confidence, select '(pc:Word|pc:Glyph)[pc:TextEquiv/@conf < 0.7]'. Or low layout confidence, '*[pc:Coords/@conf < 0.7]'.\nTo remove high pixel-to-character rate, select '*[pc:pixelarea(.) div string-length(pc:textequiv(.)) > 10000]'."
        },
        "plot": {
          "type": "boolean",
          "default": false,
          "description": "Whether to extract an image for each filtered segment and write to the output fileGrp."
        }
      }
    },
    "ocrd-command": {
      "executable": "ocrd-command",
      "description": "Bare-bones processor runs shell commands to process PAGE files",
        "steps": ["recognition/text-recognition", "recognition/font-identification", "recognition/post-correction", "layout/segmentation", "layout/analysis"],
      "categories": [],
      "input_file_grp_cardinality": [1, -1],
      "output_file_grp_cardinality": 1,
      "parameters": {
        "command": {
          "type": "string",
          "default": "cat @INFILE > @OUTFILE",
          "description": "Shell command to operate on PAGE files, with @INFILE as place-holder for the input file path(s), and @OUTFILE as place-holder for the output file path. If running on multiple input fileGrps, then @INFILE must be repeated as many times."
        }
      }
    },
    "ocrd-merge": {
      "executable": "ocrd-merge",
      "description": "Bare-bones processor merges annotations from multiple fileGrps",
        "steps": ["layout/segmentation"],
      "categories": [],
      "input_file_grp_cardinality": [1, -1],
      "output_file_grp_cardinality": 1,
      "parameters": {
      }
    }
  }
}