The processor 'segment-repir' ends wirh Exception "Exception: ocrd-segment-repair exited with non-zero return value 1" if it comes after processor 'cis-ocropy-segment' in the workflow. In a changed workflow.

In a modified workflow, where processor 'cis-ocropy-segment' is replaced by processor 'tesserocr-segment-line', the processing runs.

When using the ocrd-segment-repair, I've encountered the following Error

10:30:48.758 INFO processor.RepairSegmentation - Sanitizing region "region0071"
/home/hartwig/Projekte/ulb-dd-ocr-eval-ocrd/venv/lib/python3.6/site-packages/numpy/core/fromnumeric.py:3335: RuntimeWarning: Mean of empty slice.
  out=out, **kwargs)
/home/hartwig/Projekte/ulb-dd-ocr-eval-ocrd/venv/lib/python3.6/site-packages/numpy/core/_methods.py:161: RuntimeWarning: invalid value encountered in double_scalars
  ret = ret.dtype.type(ret / rcount)
Traceback (most recent call last):
  File "/home/hartwig/Projekte/ulb-dd-ocr-eval-ocrd/venv/bin/ocrd-segment-repair", line 8, in <module>
    sys.exit(ocrd_segment_repair())
  File "/home/hartwig/Projekte/ulb-dd-ocr-eval-ocrd/venv/lib/python3.6/site-packages/click/core.py", line 764, in __call__
    return self.main(*args, **kwargs)
  File "/home/hartwig/Projekte/ulb-dd-ocr-eval-ocrd/venv/lib/python3.6/site-packages/click/core.py", line 717, in main
    rv = self.invoke(ctx)
  File "/home/hartwig/Projekte/ulb-dd-ocr-eval-ocrd/venv/lib/python3.6/site-packages/click/core.py", line 956, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "/home/hartwig/Projekte/ulb-dd-ocr-eval-ocrd/venv/lib/python3.6/site-packages/click/core.py", line 555, in invoke
    return callback(*args, **kwargs)
  File "/home/hartwig/Projekte/ulb-dd-ocr-eval-ocrd/venv/lib/python3.6/site-packages/ocrd_segment/cli.py", line 13, in ocrd_segment_repair
    return ocrd_cli_wrap_processor(RepairSegmentation, *args, **kwargs)
  File "/home/hartwig/Projekte/ulb-dd-ocr-eval-ocrd/venv/lib/python3.6/site-packages/ocrd/decorators.py", line 60, in ocrd_cli_wrap_processor
    run_processor(processorClass, ocrd_tool, mets, workspace=workspace, **kwargs)
  File "/home/hartwig/Projekte/ulb-dd-ocr-eval-ocrd/venv/lib/python3.6/site-packages/ocrd/processor/base.py", line 57, in run_processor
    processor.process()
  File "/home/hartwig/Projekte/ulb-dd-ocr-eval-ocrd/venv/lib/python3.6/site-packages/ocrd_segment/repair.py", line 88, in process
    self.sanitize_page(page, page_id)
  File "/home/hartwig/Projekte/ulb-dd-ocr-eval-ocrd/venv/lib/python3.6/site-packages/ocrd_segment/repair.py", line 202, in sanitize_page
    scale = int(np.median(np.array(heights)))
ValueError: cannot convert float NaN to integer

The Region in question belongs to a Newspaper Digitalization.

It is possible to workaround at line 202 in repair.py (please see above) with with a check like

            _median = np.median(np.array(heights))
            if not np.isnan(_median):
                scale = int(_median)
            else:
                scale = 1

which finally yields at the same place

10:48:47.496 INFO processor.RepairSegmentation - Sanitizing region "region0071"
/home/hartwig/Projekte/ulb-dd-ocr-eval-ocrd/venv/lib/python3.6/site-packages/numpy/core/fromnumeric.py:3335: RuntimeWarning: Mean of empty slice.
  out=out, **kwargs)
/home/hartwig/Projekte/ulb-dd-ocr-eval-ocrd/venv/lib/python3.6/site-packages/numpy/core/_methods.py:161: RuntimeWarning: invalid value encountered in double_scalars
  ret = ret.dtype.type(ret / rcount)
10:48:47.699 WARNING processor.RepairSegmentation - Zero contour area in region "region0071"

but this way the processing move further.

The default output filegroup for ocrd-segment-replace-original is set to OCR-D-IMG-CROP which already exists in the majority of METS-files. Would be great to change the default value, so a user is not forced to specify it on his own (as for the other processors this is purely optional). @kba suggested, that changing the default value might actually not be necessary, if we could drop the rule for two output filegroups also for this processor.

Fixes #21 – it was not correct to use the region image/array here, because that depends on the bounding box of the region, which can be too small.

Something not covered by this is when TextLine coordinates even extrude the page Border.

Before samitization:

Regions are often too small and do not span the lines they (should) contain.

After sanitization:

Situation is not much better, although

$ ocrd-segment-repair -J
...
"sanitize": {
   "type": "boolean",
   "default": false,
   "description": "Shrink and/or expand a region in such a way that it coordinates include those of all its lines"
  }
...

Expansion does not work, is not complete.

This is supposed to be an OCR-D processor which someday will give plausibility feedback on a page's segmentation. It uses https://pypi.org/project/Shapely/ as proposed by @bertsky.

I have used ocrd-segment-extract-lines with a PAGE file, which has had some <TextLine> with a "\n" in the <Unicode>area. Unfortunately, for these lines the extraction is not done.

This example works ok:```

    <pc:TextEquiv>
      <pc:Unicode>1889</pc:Unicode>
    </pc:TextEquiv>

This example does not work:

    <pc:TextEquiv>
      <pc:Unicode>1889
      </pc:Unicode>
    </pc:TextEquiv>

==> please clarify ...

Log output:

12:06:01.982 INFO ocrd.task_sequence.run_tasks - Start processing task 'segment-repair -I OCR-D-SEG-REG -O OCR-D-SEG-REPAIR -p '{"plausibilize": true, "sanitize": false, "plausibilize_merge_min_overlap": 0.9}''
Traceback (most recent call last):
  File "/venv-20200919/bin/ocrd", line 8, in <module>
    sys.exit(cli())
  File "/venv-20200919/lib/python3.7/site-packages/click/core.py", line 829, in __call__
    return self.main(*args, **kwargs)
  File "/venv-20200919/lib/python3.7/site-packages/click/core.py", line 782, in main
    rv = self.invoke(ctx)
  File "/venv-20200919/lib/python3.7/site-packages/click/core.py", line 1259, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
  File "/venv-20200919/lib/python3.7/site-packages/click/core.py", line 1066, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "/venv-20200919/lib/python3.7/site-packages/click/core.py", line 610, in invoke
    return callback(*args, **kwargs)
  File "/venv-20200919/lib/python3.7/site-packages/ocrd/cli/process.py", line 28, in process_cli
    run_tasks(mets, log_level, page_id, tasks, overwrite)
  File "/venv-20200919/lib/python3.7/site-packages/ocrd/task_sequence.py", line 149, in run_tasks
    raise Exception("%s exited with non-zero return value %s. STDOUT:\n%s\nSTDERR:\n%s" % (task.executable, returncode, out, err))
Exception: ocrd-segment-repair exited with non-zero return value 1. STDOUT:

STDERR:
12:06:02.420 INFO processor.RepairSegmentation - INPUT FILE 0 / PHYS_0001
12:06:02.423 INFO ocrd.page_validator - Validating input file 'FILE_0001_OCR-D-SEG-REG'
12:06:02.439 INFO processor.RepairSegmentation - INPUT FILE 1 / PHYS_0002
12:06:02.440 INFO ocrd.page_validator - Validating input file 'FILE_0002_OCR-D-SEG-REG'
Traceback (most recent call last):
  File "/venv-20200919/local/sub-venv/headless-tf1/bin/ocrd-segment-repair", line 8, in <module>
    sys.exit(ocrd_segment_repair())
  File "/venv-20200919/local/sub-venv/headless-tf1/lib/python3.7/site-packages/click/core.py", line 829, in __call__
    return self.main(*args, **kwargs)
  File "/venv-20200919/local/sub-venv/headless-tf1/lib/python3.7/site-packages/click/core.py", line 782, in main
    rv = self.invoke(ctx)
  File "/venv-20200919/local/sub-venv/headless-tf1/lib/python3.7/site-packages/click/core.py", line 1066, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "/venv-20200919/local/sub-venv/headless-tf1/lib/python3.7/site-packages/click/core.py", line 610, in invoke
    return callback(*args, **kwargs)
  File "/venv-20200919/local/sub-venv/headless-tf1/lib/python3.7/site-packages/ocrd_segment/cli.py", line 16, in ocrd_segment_repair
    return ocrd_cli_wrap_processor(RepairSegmentation, *args, **kwargs)
  File "/venv-20200919/local/sub-venv/headless-tf1/lib/python3.7/site-packages/ocrd/decorators.py", line 102, in ocrd_cli_wrap_processor
    run_processor(processorClass, ocrd_tool, mets, workspace=workspace, **kwargs)
  File "/venv-20200919/local/sub-venv/headless-tf1/lib/python3.7/site-packages/ocrd/processor/helpers.py", line 69, in run_processor
    processor.process()
  File "/venv-20200919/local/sub-venv/headless-tf1/lib/python3.7/site-packages/ocrd_segment/repair.py", line 94, in process
    parents = list(set([region.parent_object_ for region in page.get_AllRegions(classes=['Text'])]))
  File "/venv-20200919/local/sub-venv/headless-tf1/lib/python3.7/site-packages/ocrd_models/ocrd_page_generateds.py", line 2905, in __hash__
    return hash(self.id)
AttributeError: 'PageType' object has no attribute 'id'

This attempts to fix problems caused by invalid polygons from ocrd-segment-repair (both in sanitize and plausibilize mode).

This taught me another lesson about what can go wrong with Shapely / numpy / PAGE interaction. To sum up:

1. Ensuring valid polygons on the input side (e.g. from OpenCV) is always necessary. The only generic way I can think of is to feed them through simplify with ever increasing tolerance until valid. EDIT2 The problem is that the result of the algorithm implemented in Shapely/GEOS depends on the starting point it picked. In pathological cases, no simplification whatsoever can be achieved. (The only thing that then helps is re-ordering...)
2. Operations like union or intersection can create collections of shapes. EDIT There are actually 2 cases here:
   1. homogeneous (MultiPolygon) – a discontiguous collection of Polygon – in which case one needs the convex hull.
   2. heterogeneous (GeometryCollection) – a collection of Polygon with Point or LineString – in which case one needs to filter out those shapes which have no intrinsic area (and then check again for the other cases)
3. Operations like union or intersection can create non-integer points, which when rounded for PAGE serialization can become invalid paths. Unfortunately, Shapely always calculates in floating point internally. So all we can do is rounding and then ensuring validity (as in 1).

Related:

• https://github.com/cisocrgroup/ocrd_cis/issues/67
• https://github.com/cisocrgroup/ocrd_cis/issues/62
• https://github.com/OCR-D/ocrd_tesserocr/issues/149
• https://github.com/OCR-D/ocrd_tesserocr/issues/151

README announces ocrd-segment-via-template and ocrd-segment-via-model – none of which are actually provided by this package. It does provide some ocrd-segment-extract-* features; these do not do any segmentation though (or I could not find out how).

The change https://github.com/OCR-D/ocrd_segment/commit/c8756272caf900febe7166f8bed5d20713f002cf depends on https://github.com/ppwwyyxx/cocoapi/pull/7, an addition of mine to the current pycocotools version 2.0.3 on PyPI. Such git URL references are allowed in requirements.txt / setuptools, but the PyPI server refuses taking such builds:

Invalid value for requires_dist. Error: Can't have direct dependency: 'pycocotools @ git+https://github.com/bertsky/pycocotools#subdirectory=PythonAPI'

@kba, do you know what to do under such circumstances?

Running make all for ocrd_all or pip install . for ocrd_segment fails on MacOS with Homebrew:

      Compiling pycocotools/_mask.pyx because it changed.
      [1/1] Cythonizing pycocotools/_mask.pyx
      /private/var/folders/wf/g2hmm5bd72v2r_p0r1smct_00000gn/T/pip-install-0xp4jh31/ocrd-fork-pycocotools_7b0159a305264f708a622a0e4daa80bd/.eggs/Cython-3.0.0a11-py3.9.egg/Cython/Compiler/Main.py:345: FutureWarning: Cython directive 'language_level' not set, using '3str' for now (Py3). This has changed from earlier releases! File: /private/var/folders/wf/g2hmm5bd72v2r_p0r1smct_00000gn/T/pip-install-0xp4jh31/ocrd-fork-pycocotools_7b0159a305264f708a622a0e4daa80bd/pycocotools/_mask.pyx
        tree = Parsing.p_module(s, pxd, full_module_name)
      building 'pycocotools._mask' extension
      creating build/common
      creating build/temp.macosx-12-arm64-cpython-39
      creating build/temp.macosx-12-arm64-cpython-39/common
      creating build/temp.macosx-12-arm64-cpython-39/pycocotools
      clang -Wno-unused-result -Wsign-compare -Wunreachable-code -fno-common -dynamic -DNDEBUG -g -fwrapv -O3 -Wall -isysroot /Library/Developer/CommandLineTools/SDKs/MacOSX12.sdk -I/OCR-D/venv-20221112/lib/python3.9/site-packages/numpy/core/include -I./common -I/OCR-D/venv-20221112/include -I/opt/homebrew/opt/[email protected]/Frameworks/Python.framework/Versions/3.9/include/python3.9 -c ../common/maskApi.c -o build/temp.macosx-12-arm64-cpython-39/../common/maskApi.o -Wno-cpp -Wno-unused-function -std=c99
      clang: error: no such file or directory: '../common/maskApi.c'
      clang: error: no input files
      error: command '/usr/bin/clang' failed with exit code 1
      [end of output]
  
  note: This error originates from a subprocess, and is likely not a problem with pip.
  ERROR: Failed building wheel for ocrd-fork-pycocotools

The segment-repair processor in the following workflow:

ocrd process \
"olena-binarize -I OCR-D-IMG -O OCR-D-BIN -P impl sauvola" \
"anybaseocr-crop -I OCR-D-BIN -O OCR-D-CROP" \
"olena-binarize -I OCR-D-CROP -O OCR-D-BIN2 -P impl kim" \
"cis-ocropy-denoise -I OCR-D-BIN2 -O OCR-D-BIN-DENOISE -P level-of-operation page" \
"cis-ocropy-deskew -I OCR-D-BIN-DENOISE -O OCR-D-BIN-DENOISE-DESKEW -P level-of-operation page" \
"tesserocr-segment-region -I OCR-D-BIN-DENOISE-DESKEW -O OCR-D-SEG-REG" \
"segment-repair -I OCR-D-SEG-REG -O OCR-D-SEG-REPAIR -P plausibilize true" \
"cis-ocropy-deskew -I OCR-D-SEG-REPAIR -O OCR-D-SEG-REG-DESKEW -P level-of-operation region" \
"cis-ocropy-clip -I OCR-D-SEG-REG-DESKEW -O OCR-D-SEG-REG-DESKEW-CLIP -P level-of-operation region" \
"tesserocr-segment-line -I OCR-D-SEG-REG-DESKEW-CLIP -O OCR-D-SEG-LINE" \
"segment-repair -I OCR-D-SEG-LINE -O OCR-D-SEG-REPAIR-LINE -P sanitize true" \
"cis-ocropy-dewarp -I OCR-D-SEG-REPAIR-LINE -O OCR-D-SEG-LINE-RESEG-DEWARP" \
"calamari-recognize -I OCR-D-SEG-LINE-RESEG-DEWARP -O OCR-D-OCR -P checkpoint_dir qurator-gt4histocr-1.0"

executed on the DEFAULT file group inside this workspace: https://content.staatsbibliothek-berlin.de/dc/PPN631277528.mets.xml

produces the following error:

  12:45:52.522 INFO processor.RepairSegmentation - INPUT FILE 0 / PHYS_0001
  12:45:52.524 INFO ocrd.page_validator.validate - Validating input file 'FILE_0001_OCR-D-SEG-LINE'
  12:45:52.652 INFO processor.RepairSegmentation - INPUT FILE 1 / PHYS_0002
  12:45:52.654 INFO ocrd.page_validator.validate - Validating input file 'FILE_0002_OCR-D-SEG-LINE'
  12:45:52.776 INFO processor.RepairSegmentation - INPUT FILE 2 / PHYS_0003
  12:45:52.777 INFO ocrd.page_validator.validate - Validating input file 'FILE_0003_OCR-D-SEG-LINE'
  12:45:52.912 INFO processor.RepairSegmentation - INPUT FILE 3 / PHYS_0004
  12:45:52.914 INFO ocrd.page_validator.validate - Validating input file 'FILE_0004_OCR-D-SEG-LINE'
  12:45:53.017 INFO processor.RepairSegmentation - INPUT FILE 4 / PHYS_0005
  12:45:53.019 INFO ocrd.page_validator.validate - Validating input file 'FILE_0005_OCR-D-SEG-LINE'
  12:45:53.026 WARNING processor.RepairSegmentation - Fixed CoordinateValidityError for SeparatorRegion 'region0011'
  12:45:53.027 WARNING processor.RepairSegmentation - Fixed CoordinateValidityError for SeparatorRegion 'region0012'
  12:45:53.119 WARNING processor.RepairSegmentation - Zero contour area in region "region0000"
  12:45:53.730 WARNING processor.RepairSegmentation - Zero contour area in region "region0011"
  12:45:53.734 WARNING processor.RepairSegmentation - Zero contour area in region "region0012"
  12:45:54.609 INFO processor.RepairSegmentation - INPUT FILE 5 / PHYS_0006
  12:45:54.610 INFO ocrd.page_validator.validate - Validating input file 'FILE_0006_OCR-D-SEG-LINE'
  12:45:54.708 INFO processor.RepairSegmentation - INPUT FILE 6 / PHYS_0007
  12:45:54.710 INFO ocrd.page_validator.validate - Validating input file 'FILE_0007_OCR-D-SEG-LINE'
  12:45:54.812 WARNING processor.RepairSegmentation - Zero contour area in region "region0003"
  12:45:55.186 ERROR shapely.geos - TopologyException: side location conflict at 262 1071. This can occur if the input geometry is invalid.
  Traceback (most recent call last):
    File "/home/mm/venv37-ocrd/sub-venv/headless-tf1/bin/ocrd-segment-repair", line 8, in <module>
      sys.exit(ocrd_segment_repair())
    File "/home/mm/venv37-ocrd/sub-venv/headless-tf1/lib/python3.7/site-packages/click/core.py", line 1130, in __call__
      return self.main(*args, **kwargs)
    File "/home/mm/venv37-ocrd/sub-venv/headless-tf1/lib/python3.7/site-packages/click/core.py", line 1055, in main
      rv = self.invoke(ctx)
    File "/home/mm/venv37-ocrd/sub-venv/headless-tf1/lib/python3.7/site-packages/click/core.py", line 1404, in invoke
      return ctx.invoke(self.callback, **ctx.params)
    File "/home/mm/venv37-ocrd/sub-venv/headless-tf1/lib/python3.7/site-packages/click/core.py", line 760, in invoke
      return __callback(*args, **kwargs)
    File "/home/mm/venv37-ocrd/sub-venv/headless-tf1/lib/python3.7/site-packages/ocrd_segment/cli.py", line 21, in ocrd_segment_repair
      return ocrd_cli_wrap_processor(RepairSegmentation, *args, **kwargs)
    File "/home/mm/venv37-ocrd/sub-venv/headless-tf1/lib/python3.7/site-packages/ocrd/decorators/__init__.py", line 108, in ocrd_cli_wrap_processor
      run_processor(processorClass, ocrd_tool, mets, workspace=workspace, **kwargs)
    File "/home/mm/venv37-ocrd/sub-venv/headless-tf1/lib/python3.7/site-packages/ocrd/processor/helpers.py", line 88, in run_processor
      processor.process()
    File "/home/mm/venv37-ocrd/sub-venv/headless-tf1/lib/python3.7/site-packages/ocrd_segment/repair.py", line 188, in process
      padding=self.parameter['sanitize_padding'])
    File "/home/mm/venv37-ocrd/sub-venv/headless-tf1/lib/python3.7/site-packages/ocrd_segment/repair.py", line 559, in shrink_regions
      if len(contour) >= 3], scale=scale)
    File "/home/mm/venv37-ocrd/sub-venv/headless-tf1/lib/python3.7/site-packages/ocrd_segment/project.py", line 179, in join_polygons
      jointp = unary_union(polygons)
    File "/home/mm/venv37-ocrd/sub-venv/headless-tf1/lib/python3.7/site-packages/shapely/ops.py", line 161, in unary_union
      return geom_factory(lgeos.methods['unary_union'](collection))
    File "/home/mm/venv37-ocrd/sub-venv/headless-tf1/lib/python3.7/site-packages/shapely/geometry/base.py", line 73, in geom_factory
      raise ValueError("No Shapely geometry can be created from null value")
  ValueError: No Shapely geometry can be created from null value

This is the input image:

Hi, I think I have found a bug in ocrd-segment-extract-lines: I cannot prove to 100%, but I think I see my environment, that the lines are not extracted (no images are created), in case a line is somehow graphically (concerning the coordinates) within another line of the same region. I extract only images in this case using this command:

ocrd-segment-extract-lines -I $infolder -O $extractLineImagesFolder  -P  output-types '[]' -P min-line-length 0 -P min-line-width 5 -P min-line-height 5

Page-Extract: Here the line TR-15_line0002 was not extracted:

    <pc:TextRegion id="TR-15" orientation="0.">
      <pc:AlternativeImage filename="OCR-D-REG-VL-BL/OCR-D-REG-VL-BL_4749_007817786_00183_TR-15.IMG-DESKEW.png" comments=",binarized,deskewed,verticallinesremoved" />
      <pc:Coords points="237,383 237,438 443,438 443,383" />
      <pc:TextLine id="TR-15_line0001">
        <pc:Coords points="237,438 237,383 239,383 253,391 311,391 320,383 349,383 357,390 365,383 384,383 402,391 419,383 427,383 430,418 428,438 302,438 298,435 289,435 284,438" />
        <pc:Baseline points="227,415 430,418" />
      </pc:TextLine>
      <pc:TextLine id="TR-15_line0003">
        <pc:Coords points="261,438 269,433 274,433 295,438" />
        <pc:Baseline points="254,475 295,475" />
      </pc:TextLine>
      <pc:TextLine id="TR-15_line0002">
        <pc:Coords points="385,438 388,435 388,434 409,434 409,438" />
        <pc:Baseline points="343,478 412,475" />
      </pc:TextLine>
    </pc:TextRegion>

Logfile content for this case:

2022-08-11_14-21-13-extractlines.log-2022-08-11 14:21:31.189 WARNING processor.ExtractLines - Line 'TR-14_line0001' contains no text content
2022-08-11_14-21-13-extractlines.log-2022-08-11 14:21:31.201 INFO ocrd.workspace.save_image_file - created file ID: OCR-D-SEG-LINE-CCS-IMG-BL-4749_007817786_00183_TR-14_TR-14_line0001.bin, file_grp: OCR-D-SEG-LINE-CCS-IMG-BL, path: OCR-D-SEG-LINE-CCS-IMG-BL/OCR-D-SEG-LINE-CCS-IMG-BL-4749_007817786_00183_TR-14_TR-14_line0001.bin.png
2022-08-11_14-21-13-extractlines.log-2022-08-11 14:21:31.242 WARNING processor.ExtractLines - Line 'TR-15_line0001' contains no text content
2022-08-11_14-21-13-extractlines.log:2022-08-11 14:21:31.255 INFO ocrd.workspace.save_image_file - created file ID: OCR-D-SEG-LINE-CCS-IMG-BL-4749_007817786_00183_TR-15_TR-15_line0001.bin, file_grp: OCR-D-SEG-LINE-CCS-IMG-BL, path: OCR-D-SEG-LINE-CCS-IMG-BL/OCR-D-SEG-LINE-CCS-IMG-BL-4749_007817786_00183_TR-15_TR-15_line0001.bin.png
2022-08-11_14-21-13-extractlines.log-2022-08-11 14:21:31.256 WARNING processor.ExtractLines - Line 'TR-15_line0003' contains no text content
2022-08-11_14-21-13-extractlines.log-2022-08-11 14:21:31.267 INFO ocrd.workspace.save_image_file - created file ID: OCR-D-SEG-LINE-CCS-IMG-BL-4749_007817786_00183_TR-15_TR-15_line0003.bin, file_grp: OCR-D-SEG-LINE-CCS-IMG-BL, path: OCR-D-SEG-LINE-CCS-IMG-BL/OCR-D-SEG-LINE-CCS-IMG-BL-4749_007817786_00183_TR-15_TR-15_line0003.bin.png
2022-08-11_14-21-13-extractlines.log-2022-08-11 14:21:31.268 WARNING processor.ExtractLines - Line 'TR-15_line0002' contains no text content
2022-08-11_14-21-13-extractlines.log-2022-08-11 14:21:31.311 WARNING processor.ExtractLines - Line 'TR-16_line0001' contains no text content
2022-08-11_14-21-13-extractlines.log-2022-08-11 14:21:31.348 INFO ocrd.workspace.save_image_file - created file ID: OCR-D-SEG-LINE-CCS-IMG-BL-4749_007817786_00183_TR-16_TR-16_line0001.bin, file_grp: OCR-D-SEG-LINE-CCS-IMG-BL, path: OCR-D-SEG-LINE-CCS-IMG-BL/OCR-D-SEG-LINE-CCS-IMG-BL-4749_007817786_00183_TR-16_TR-16_line0001.bin.png

Version 0.1.20, ocrd/core 2.33.0

I have a PAGE file, which does not have any real content - like this:

    <pc:Page imageFilename="OCR-D-IMG/0038_IMAGE000918_00001.tif" imageWidth="1420" imageHeight="2313" orientation="0.">
        <pc:AlternativeImage filename="OCR-D-BIN/OCR-D-BIN_0038_IMAGE000918_00001.IMG-BIN.png" comments=",binarized"/>
        <pc:TextRegion id="TR-1" orientation="0.">
            <pc:Coords points=""/>
        </pc:TextRegion>
    </pc:Page>

If I call ocrd-segment-extract-lines, I get an expection like this:

09:19:19.733 DEBUG ocrd.workspace.image_from_page - page 'P_0038_IMAGE000918_00001' has  orientation=0 skew=0.00
09:19:19.733 DEBUG ocrd.workspace.image_from_page - Using AlternativeImage 1 {'', 'binarized'} for page 'P_0038_IMAGE000918_00001'
09:19:19.734 DEBUG ocrd.workspace.download_file - download_file <OcrdFile fileGrp=OCR-D-BIN ID=OCR-D-BIN_0038_IMAGE000918_00001.IMG-BIN, mimetype=image/png, url=OCR-D-BIN/OCR-D-BIN_0038_IMAGE000918_00001.IMG-BIN.png, local_filename=OCR-D-BIN/OCR-D-BIN_0038_IMAGE000918_00001.IMG-BIN.png]/>  [_recursion_count=0]
09:19:19.735 DEBUG PIL.PngImagePlugin - STREAM b'IHDR' 16 13
09:19:19.735 DEBUG PIL.PngImagePlugin - STREAM b'IDAT' 41 65536
Traceback (most recent call last):
  File "/home/ocrdadmin/ocrd_all/venv/sub-venv/headless-tf1/bin/ocrd-segment-extract-lines", line 8, in <module>
    sys.exit(ocrd_segment_extract_lines())
  File "/home/ocrdadmin/ocrd_all/venv/sub-venv/headless-tf1/lib/python3.6/site-packages/click/core.py", line 1128, in __call__
    return self.main(*args, **kwargs)
  File "/home/ocrdadmin/ocrd_all/venv/sub-venv/headless-tf1/lib/python3.6/site-packages/click/core.py", line 1053, in main
    rv = self.invoke(ctx)
  File "/home/ocrdadmin/ocrd_all/venv/sub-venv/headless-tf1/lib/python3.6/site-packages/click/core.py", line 1395, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "/home/ocrdadmin/ocrd_all/venv/sub-venv/headless-tf1/lib/python3.6/site-packages/click/core.py", line 754, in invoke
    return __callback(*args, **kwargs)
  File "/home/ocrdadmin/ocrd_all/venv/sub-venv/headless-tf1/lib/python3.6/site-packages/ocrd_segment/cli.py", line 65, in ocrd_segment_extract_lines
    return ocrd_cli_wrap_processor(ExtractLines, *args, **kwargs)
  File "/home/ocrdadmin/ocrd_all/venv/sub-venv/headless-tf1/lib/python3.6/site-packages/ocrd/decorators/__init__.py", line 88, in ocrd_cli_wrap_processor
    run_processor(processorClass, ocrd_tool, mets, workspace=workspace, **kwargs)
  File "/home/ocrdadmin/ocrd_all/venv/sub-venv/headless-tf1/lib/python3.6/site-packages/ocrd/processor/helpers.py", line 88, in run_processor
    processor.process()
  File "/home/ocrdadmin/ocrd_all/venv/sub-venv/headless-tf1/lib/python3.6/site-packages/ocrd_segment/extract_lines.py", line 171, in process
    transparency=self.parameter['transparency'])
  File "/home/ocrdadmin/ocrd_all/venv/sub-venv/headless-tf1/lib/python3.6/site-packages/ocrd/workspace.py", line 829, in image_from_segment
    fill=fill, transparency=transparency)
  File "/home/ocrdadmin/ocrd_all/venv/sub-venv/headless-tf1/lib/python3.6/site-packages/ocrd/workspace.py", line 1012, in _crop
    segment_polygon = coordinates_of_segment(segment, parent_image, parent_coords)
  File "/home/ocrdadmin/ocrd_all/venv/sub-venv/headless-tf1/lib/python3.6/site-packages/ocrd_utils/image.py", line 136, in coordinates_of_segment
    polygon = np.array(polygon_from_points(segment.get_Coords().points))
  File "/home/ocrdadmin/ocrd_all/venv/sub-venv/headless-tf1/lib/python3.6/site-packages/ocrd_utils/image.py", line 148, in polygon_from_points
    polygon.append([float(x_y[0]), float(x_y[1])])
ValueError: could not convert string to float: 

My expection would be, that this PAGE file simply would be ignored. --> please, clarify ...

If I understand correctly the idea behind these metrics are taken from "rethinking semantic segmentation evaluation" paper, but could you explain to me how could I obtain AP,TPs,FPs,FNs for instance segmentation task?

Originally posted by @andreaceruti in https://github.com/cocodataset/cocoapi/issues/564#issuecomment-1064223428

The multi-match overlap algorithm (necessary to calculate over- and undersegmentation) still has a glitch: it will create fake/redundant pairings if either side has a segmentation that already overlaps locally. For example, take a page with a GraphicRegion overlapping multiple TextRegions, and evaluate that against itself: the matching will not only produce the 1:1 pairs, but also other matches. That's probably not what we want.