tesseract训练脚本

@echo *********************************************

@echo 欢迎关注公众号：小鸡炖技术，本脚本由小T提供

@echo *********************************************

@pause

@set name=chi_my

@set font=font

:: 中文写chi_sim,英文或数字填eng

@set lang=chi_sim

:: tif的名字

:: chi_my.font.exp0.tif

@if not exist %name%.%font%.exp0.box ( goto LABLE_MAKEBOX ) else ( goto LABLE_TRIAN )

:LABLE_MAKEBOX

@echo -------step 1 start -----------

tesseract %name%.%font%.exp0.tif %name%.%font%.exp0 -l %lang% --psm 7 batch.nochop makebox

@echo -------step 1 stop-----------

exit

:LABLE_TRIAN

@if exist *.traineddata ( del .\*.traineddata /f /s /q /a )

@echo -------step 2 start -----------

echo "%font% 0 0 0 0 0"> %font%_properties

@echo -------step 2 stop -----------

@echo -------step 3 start -----------

tesseract %name%.%font%.exp0.tif %name%.%font%.exp0 nobatch box.train

@echo -------step 3 stop -----------

@echo -------step 4 start -----------

unicharset_extractor %name%.%font%.exp0.box

@echo -------step 4 stop -----------

@echo -------step 5 start -----------

mftraining -F font_properties -U unicharset -O %name%.unicharset %name%.%font%.exp0.tr

@echo -------step 5 stop -----------

@echo -------step 6 start -----------

cntraining %name%.%font%.exp0.tr

@echo -------step 6 stop -----------

@echo -------step 7 start -----------

@rename normproto %name%.normproto

@rename inttemp %name%.inttemp

@rename pffmtable %name%.pffmtable

@rename shapetable %name%.shapetable

@echo -------step 7 stop -----------

@echo -------step 8 start -----------

combine_tessdata %name%.

@echo -------step 8 stop -----------

@echo -------step 9 start -----------

del .\*.tr /f /s /q /a

del .\*.normproto /f /s /q /a

del .\*.inttemp /f /s /q /a

del .\*.pffmtable /f /s /q /a

del .\*.shapetable /f /s /q /a

del .\*.unicharset /f /s /q /a

del .\unicharset /f /s /q /a

del .\font_properties /f /s /q /a

@echo -------step 9 stop -----------

@echo ******************************************

@echo 恭喜你，训练结束！

@echo 欢迎关注公众号：小鸡炖技术，本脚本由小T提供

@echo ******************************************

、