def runTest(self):
resolver = Resolver()
workspace = resolver.workspace_from_url(METS_HEROLD_SMALL,
directory=WORKSPACE_DIR)
TesserocrSegmentRegion(workspace,
input_file_grp="INPUT",
output_file_grp="OCR-D-SEG-BLOCK").process()
workspace.save_mets()
def runTest(self):
resolver = Resolver()
workspace = resolver.workspace_from_url(METS_HEROLD_SMALL,
dst_dir=WORKSPACE_DIR)
TesserocrSegmentRegion(workspace,
input_file_grp="OCR-D-IMG",
output_file_grp="OCR-D-SEG-BLOCK").process()
TesserocrSegmentLine(workspace,
input_file_grp="OCR-D-SEG-BLOCK",
output_file_grp="OCR-D-SEG-LINE").process()
TesserocrSegmentWord(workspace,
input_file_grp="OCR-D-SEG-LINE",
output_file_grp="OCR-D-SEG-WORD").process()
workspace.save_mets()
def runTest(self):
resolver = Resolver()
# workspace = resolver.workspace_from_url(assets.url_of('SBB0000F29300010000/mets_one_file.xml'), directory=WORKSPACE_DIR)
workspace = resolver.workspace_from_url(
assets.url_of('kant_aufklaerung_1784-binarized/mets.xml'),
directory=WORKSPACE_DIR)
TesserocrSegmentRegion(workspace,
input_file_grp="OCR-D-IMG",
output_file_grp="OCR-D-SEG-BLOCK").process()
TesserocrSegmentLine(workspace,
input_file_grp="OCR-D-SEG-BLOCK",
output_file_grp="OCR-D-SEG-LINE").process()
TesserocrSegmentWord(workspace,
input_file_grp="OCR-D-SEG-LINE",
output_file_grp="OCR-D-SEG-WORD").process()
workspace.save_mets()
def runTest(self):
resolver = Resolver(cache_enabled=True)
# workspace = resolver.workspace_from_url(assets.url_of('SBB0000F29300010000/mets_one_file.xml'), directory=WORKSPACE_DIR)
workspace = resolver.workspace_from_url(assets.url_of(
'kant_aufklaerung_1784-page-block-line-word/mets.xml'),
directory=WORKSPACE_DIR)
TesserocrSegmentRegion(workspace,
input_file_grp="OCR-D-IMG",
output_file_grp="OCR-D-SEG-BLOCK").process()
workspace.save_mets()
TesserocrSegmentLine(workspace,
input_file_grp="OCR-D-SEG-BLOCK",
output_file_grp="OCR-D-SEG-LINE").process()
workspace.save_mets()
TesserocrRecognize(workspace,
input_file_grp="OCR-D-SEG-LINE",
output_file_grp="OCR-D-OCR-TESS",
parameter={
'textequiv_level': 'word'
}).process()
workspace.save_mets()
def runTest(self):
resolver = Resolver()
workspace = resolver.workspace_from_url(METS_HEROLD_SMALL,
dst_dir=WORKSPACE_DIR)
TesserocrSegmentRegion(workspace,
input_file_grp="OCR-D-IMG",
output_file_grp="OCR-D-SEG-BLOCK").process()
workspace.save_mets()
TesserocrSegmentLine(workspace,
input_file_grp="OCR-D-SEG-BLOCK",
output_file_grp="OCR-D-SEG-LINE").process()
workspace.save_mets()
TesserocrRecognize(
workspace,
input_file_grp="OCR-D-SEG-LINE",
output_file_grp="OCR-D-OCR-TESS",
parameter={
'textequiv_level': 'line'
} # add dep tesseract-ocr-script-frak: , 'model': 'Fraktur'
).process()
workspace.save_mets()
TesserocrSegmentWord(workspace,
input_file_grp="OCR-D-SEG-LINE",
output_file_grp="OCR-D-SEG-WORD").process()
workspace.save_mets()
TesserocrRecognize(
workspace,
input_file_grp="OCR-D-SEG-WORD",
output_file_grp="OCR-D-OCR-TESS-W2C",
parameter={
'textequiv_level': 'glyph'
} # add dep tesseract-ocr-script-frak: , 'model': 'Fraktur'}
).process()
workspace.save_mets()
def runTest(self):
resolver = Resolver()
# workspace = resolver.workspace_from_url(assets.url_of('SBB0000F29300010000/mets_one_file.xml'), directory=WORKSPACE_DIR)
workspace = resolver.workspace_from_url(assets.url_of(
'kant_aufklaerung_1784-page-block-line-word/mets.xml'),
directory=WORKSPACE_DIR)
TesserocrSegmentRegion(workspace,
input_file_grp="OCR-D-IMG",
output_file_grp="OCR-D-SEG-BLOCK").process()
workspace.save_mets()
TesserocrSegmentLine(workspace,
input_file_grp="OCR-D-SEG-BLOCK",
output_file_grp="OCR-D-SEG-LINE").process()
workspace.save_mets()
TesserocrRecognize(
workspace,
input_file_grp="OCR-D-SEG-LINE",
output_file_grp="OCR-D-OCR-TESS",
parameter={
'textequiv_level': 'line'
} # add dep tesseract-ocr-script-frak: , 'model': 'Fraktur'
).process()
workspace.save_mets()
TesserocrSegmentWord(workspace,
input_file_grp="OCR-D-SEG-LINE",
output_file_grp="OCR-D-SEG-WORD").process()
workspace.save_mets()
TesserocrRecognize(
workspace,
input_file_grp="OCR-D-SEG-WORD",
output_file_grp="OCR-D-OCR-TESS-W2C",
parameter={
'textequiv_level': 'glyph'
} # add dep tesseract-ocr-script-frak: , 'model': 'Fraktur'}
).process()
workspace.save_mets()
def runTest(self):
resolver = Resolver(cache_enabled=True)
workspace = resolver.workspace_from_url(
assets.url_of('SBB0000F29300010000/mets_one_file.xml'))
TesserocrSegmentRegion(workspace).process()
workspace.save_mets()
