Tesseract 4 beta support - Orientation detection failure
Hello,
Would it be possible to have again the ability to deactivate orientation detection ?
Rationale:
tesseract method is supposed to be fast and reliable. In reality, several bugs in tesseract, pyocr, version incompatibilities, etc. make this mandatory feature a burden rather than a nice bonus while at the same time 99% of paper scanned is done on the same orientation.
I'm writing this while I'm scanning a 50+ page document, that takes 40 seconds per page to perform OCR.
While tesseract is installed and my installation worked some times ago, now I have this message:
INFO paperwork.frontend.mainwindow.scan Failed to use OCR tool heuristic for orientation detection: (-1, "No script found in image (read_params_file: Can't open 0\nread_params_file: Can't open -oem\nread_params_file: Can't open 0\nread_params_file: Can't open l\nread_params_file: Can't open fra\nReéférences a rappeler...
The success rate of the spell checking method is close to 10% so far.
Please make orientation optional again.
Edited by Jerome Flesch