ValueError in PP_structureV3- all the input array dimensions except for the concatenation axis must match exactly, but along dimension 1, the array at index 0 has size 0 and the array at index 1 has size 4

[devenus1]

I am encountering a "ValueError: all the input array dimensions except for the concatenation axis must match exactly, but along dimension 1, the array at index 0 has size 0 and the array at index 1 has size 4" while running using the PPStructureV3 pipeline.

Code Used:

from paddleocr import PPStructureV3

pipeline = PPStructureV3(
    use_seal_recognition = False,
    use_chart_recognition = False,
    use_formula_recognition = False,
    use_doc_unwarping = False,
    use_doc_orientation_classify = False,
    use_region_detection = False,
    use_table_recognition = True,
)

output = pipeline.predict(
    input="processed_images/page_4.png",
    use_doc_orientation_classify=False,
    use_doc_unwarping=False,
)

Error :

  File "/home/ubuntu/paddle-error/.venv/lib/python3.10/site-packages/paddleocr/_pipelines/pp_structurev3.py", line 206, in predict
    return list(
  File "/home/ubuntu/paddle-error/.venv/lib/python3.10/site-packages/paddlex/inference/pipelines/_parallel.py", line 129, in predict
    yield from self._pipeline.predict(
  File "/home/ubuntu/paddle-error/.venv/lib/python3.10/site-packages/paddlex/inference/pipelines/layout_parsing/pipeline_v2.py", line 1282, in predict
    parsing_res_list = self.get_layout_parsing_res(
  File "/home/ubuntu/paddle-error/.venv/lib/python3.10/site-packages/paddlex/inference/pipelines/layout_parsing/pipeline_v2.py", line 857, in get_layout_parsing_res
    self.standardized_data(
  File "/home/ubuntu/paddle-error/.venv/lib/python3.10/site-packages/paddlex/inference/pipelines/layout_parsing/pipeline_v2.py", line 464, in standardized_data
    overall_ocr_res["rec_boxes"] = np.vstack(
  File "<__array_function__ internals>", line 200, in vstack
  File "/home/ubuntu/paddle-error/.venv/lib/python3.10/site-packages/numpy/core/shape_base.py", line 296, in vstack
    return _nx.concatenate(arrs, 0, dtype=dtype, casting=casting)
  File "<__array_function__ internals>", line 200, in concatenate
ValueError: all the input array dimensions except for the concatenation axis must match exactly, but along dimension 1, the array at index 0 has size 0 and the array at index 1 has size 4

Issues and Observation

• The error occurs intermittently but always for the same image.
• The images appear to be correct (no obvious issues, also I have tested similar images which are returning results).
• This issue happens for only a few images, and it seems random, but it happens consistently for the same image every time.

How to avoid/correct this error?

[changdazhou]

This bug has been resolved in the most recent update. Please test the latest version.