OCR issues ----- cv2.error: OpenCV(4.11.0) :-1: error: (-5:Bad argument) in function 'fillPoly'

[YYH211]

Traceback (most recent call last): File "/opt/conda/envs/marker2/bin/marker_single", line 8, in sys.exit(convert_single_cli()) File "/opt/conda/envs/marker2/lib/python3.10/site-packages/click/core.py", line 1161, in call return self.main(*args, **kwargs) File "/opt/conda/envs/marker2/lib/python3.10/site-packages/click/core.py", line 1082, in main rv = self.invoke(ctx) File "/opt/conda/envs/marker2/lib/python3.10/site-packages/click/core.py", line 1443, in invoke return ctx.invoke(self.callback, **ctx.params) File "/opt/conda/envs/marker2/lib/python3.10/site-packages/click/core.py", line 788, in invoke return __callback(*args, **kwargs) File "/opt/conda/envs/marker2/lib/python3.10/site-packages/marker/scripts/convert_single.py", line 35, in convert_single_cli rendered = converter(fpath) File "/opt/conda/envs/marker2/lib/python3.10/site-packages/marker/converters/pdf.py", line 154, in call document = self.build_document(filepath) File "/opt/conda/envs/marker2/lib/python3.10/site-packages/marker/converters/pdf.py", line 149, in build_document processor(document) File "/opt/conda/envs/marker2/lib/python3.10/site-packages/marker/processors/table.py", line 99, in call self.assign_ocr_lines(ocr_blocks) # Handle tables where OCR is needed File "/opt/conda/envs/marker2/lib/python3.10/site-packages/marker/processors/table.py", line 382, in assign_ocr_lines ocr_results: List[OCRResult] = self.recognition_model( File "/opt/conda/envs/marker2/lib/python3.10/site-packages/surya/recognition/init.py", line 55, in call flat = self.detect_and_slice_bboxes( File "/opt/conda/envs/marker2/lib/python3.10/site-packages/surya/recognition/init.py", line 132, in detect_and_slice_bboxes slices = slice_polys_from_image(image, polygons) File "/opt/conda/envs/marker2/lib/python3.10/site-packages/surya/input/processing.py", line 43, in slice_polys_from_image lines.append(slice_and_pad_poly(image_array, poly)) File "/opt/conda/envs/marker2/lib/python3.10/site-packages/surya/input/processing.py", line 58, in slice_and_pad_poly cv2.fillPoly(mask, [np.int32(coordinates)], 1) cv2.error: OpenCV(4.11.0) :-1: error: (-5:Bad argument) in function 'fillPoly'

Overload resolution failed:

• Layout of the output array img is incompatible with cv::Mat
• Expected Ptrcv::UMat for argument 'img'

i use this script: marker_single ../maker-master/file/ocr_test.pdf --output_dir ./marker2-result --strip_existing_ocr --languages "zh"

here is the file:ocr_test.pdf

[hanruiyao163]

same issue

[mzanm]

I'm also facing the same issue as well. I've tried Python 3.11 and Python 3.12, no difference. Using version 1.6.

[YYH211]

I'm also facing the same issue as well. I've tried Python 3.11 and Python 3.12, no difference. Using version 1.6.

hello ，what is mean 1.6？ python 3.16？

[mzanm]

I'm also facing the same issue as well. I've tried Python 3.11 and Python 3.12, no difference. Using version 1.6.

hello ，what is mean 1.6？ python 3.16？

Hello. No, I meant version 1.6.0 or 1.6.1 of marker-pdf itself.

I've found a fix, you can install version 1.5.5 and it doesn't have this issue via pip install marker-pdf==1.5.5

I hope that helps until it's resolved.

[tarun-menta]

@YYH211 @mzanm @hanruiyao163 This issue is now fixed in the new release (https://github.com/VikParuchuri/marker/releases/tag/v1.6.2). Thanks for the issue report that helped us find the bug!