from-masks: pick default colordict from parameter, omit initial # sig…

…n from colorspecs
OCR-D · Feb 6, 2021 · 980b07d · 980b07d
1 parent b8ea41a
commit 980b07d
Show file tree

Hide file tree

Showing 2 changed files with 56 additions and 10 deletions.
diff --git a/ocrd_segment/import_image_segmentation.py b/ocrd_segment/import_image_segmentation.py
@@ -42,7 +42,6 @@
 from ocrd import Processor
 
 from .config import OCRD_TOOL
-from .extract_pages import CLASSES
 
 TOOL = 'ocrd-segment-from-masks'
 
@@ -71,11 +70,6 @@ def process(self):
         assert_file_grp_cardinality(self.output_file_grp, 1)
 
         colordict = self.parameter['colordict']
-        if not colordict:
-            LOG.info('Using default PAGE colordict')
-            colordict = dict(('#' + col, name)
-                             for name, col in CLASSES.items()
-                             if name)
         typedict = {"TextRegion": TextTypeSimpleType,
                     "GraphicRegion": GraphicsTypeSimpleType,
                     "ChartType": ChartTypeSimpleType}
@@ -96,9 +90,9 @@ def process(self):
                 segmentation_pil = Image.open(segmentation_filename)
             has_alpha = segmentation_pil.mode == 'RGBA'
             if has_alpha:
-                colorformat = "#%08X"
+                colorformat = "%08X"
             else:
-                colorformat = "#%06X"
+                colorformat = "%06X"
                 if segmentation_pil.mode != 'RGB':
                     segmentation_pil = segmentation_pil.convert('RGB')
             # convert to array

diff --git a/ocrd_segment/ocrd-tool.json b/ocrd_segment/ocrd-tool.json
@@ -48,8 +48,60 @@
       "parameters": {
         "colordict": {
           "type": "object",
-          "default": {},
-          "description": "Mapping from color values in the input masks to region types to annotate; color must be encoded hexadecimal (e.g. '#00FF00'); region type equals the element name in PAGE-XML, optionally followed by a colon and a subtype (e.g. 'TextRegion:paragraph'; unmapped colors will be ignored (i.e. treated as background)). Cf. output of ocrd-segment-extract-pages for an example (this is also the default)."
+          "default": {
+            "FFFFFF00": "",
+            "FFFFFFFF": "Border",
+            "8B4513FF": "TableRegion",
+            "4682B4FF": "AdvertRegion",
+            "FF8C00FF": "ChemRegion",
+            "9400D3FF": "MusicRegion",
+            "9ACDD2FF": "MapRegion",
+            "0000FFFF": "TextRegion",
+            "0000FFFA": "TextRegion:paragraph",
+            "0000FFF5": "TextRegion:heading",
+            "0000FFF0": "TextRegion:caption",
+            "0000FFEB": "TextRegion:header",
+            "0000FFE6": "TextRegion:footer",
+            "0000FFE1": "TextRegion:page-number",
+            "0000FFDC": "TextRegion:drop-capital",
+            "0000FFD7": "TextRegion:credit",
+            "0000FFD2": "TextRegion:floating",
+            "0000FFCD": "TextRegion:signature-mark",
+            "0000FFC8": "TextRegion:catch-word",
+            "0000FFC3": "TextRegion:marginalia",
+            "0000FFBE": "TextRegion:footnote",
+            "0000FFB9": "TextRegion:footnote-continued",
+            "0000FFB4": "TextRegion:endnote",
+            "0000FFAF": "TextRegion:TOC-entry",
+            "0000FFA5": "TextRegion:list-label",
+            "0000FFA0": "TextRegion:other",
+            "800080FF": "ChartRegion",
+            "800080FA": "ChartRegion:bar",
+            "800080F5": "ChartRegion:line",
+            "800080F0": "ChartRegion:pie",
+            "800080EB": "ChartRegion:scatter",
+            "800080E6": "ChartRegion:surface",
+            "800080E1": "ChartRegion:other",
+            "008000FF": "GraphicRegion",
+            "008000FA": "GraphicRegion:logo",
+            "008000F0": "GraphicRegion:letterhead",
+            "008000EB": "GraphicRegion:decoration",
+            "008000E6": "GraphicRegion:frame",
+            "008000E1": "GraphicRegion:handwritten-annotation",
+            "008000DC": "GraphicRegion:stamp",
+            "008000D7": "GraphicRegion:signature",
+            "008000D2": "GraphicRegion:barcode",
+            "008000CD": "GraphicRegion:paper-grow",
+            "008000C8": "GraphicRegion:punch-hole",
+            "008000C3": "GraphicRegion:other",
+            "00CED1FF": "ImageRegion",
+            "B8860BFF": "LineDrawingRegion",
+            "00BFFFFF": "MathsRegion",
+            "FF0000FF": "NoiseRegion",
+            "FF00FFFF": "SeparatorRegion",
+            "646464FF": "UnknownRegion",
+            "637C81FF": "CustomRegion"},
+          "description": "Mapping from color values in the input masks to region types to annotate; color must be encoded hexadecimal (e.g. '00FF00'); region type equals the element name in PAGE-XML, optionally followed by a colon and a subtype (e.g. 'TextRegion:paragraph'; unmapped colors will be ignored (i.e. treated as background)). Default is PageViewer color scheme. Cf. colordict.json output and colordict parameter of ocrd-segment-extract-pages."
         }
       }
     },