How I can specify specific segmentation mode for tesseract that Layoutlm should use?
#19
by
anemilentsau
- opened
Hi,
I am having an issue that default segmentation model for tesseract fails to extract properly the text from my document that leads to Layoutlm being incapable to provide correct answers. The proper segmentation mode shall be --psm 6. Is there a way to specify the tesseract segmentation mode that nlp pipeline for the question answering should use?