I'm pretty sure you have to have a don't for lstm training. When I trained tesseract 4 for hand writing, I used a font that was based on handwriting to fulfill tesseract's requirement for at least one font.
On Wed, Feb 6, 2019, 11:10 PM <[email protected] wrote: > Thanks for your response, Since these are handwritten digits I don't have > font data and what I'm having is cropped image blocks and I prepared some > .gt.txt files. Is it possible to do lstm training without font data? > > On Tuesday, February 5, 2019 at 1:17:27 AM UTC+5:30, Lorenzo Blz wrote: >> >> >> To use ocrd you need to prepare image files and txt files with the same >> name but different extension. >> For example: >> >> sample1.png >> sample1.gt.txt >> >> The gt.txt is a simple text file containing the correct text, 145, for >> example. >> >> The images must be cropped with no border or just a couple of pixels. >> Text height should be about 30/40px. Try different options to see what >> works best. >> >> To recognize numbers ONLY you also need to replaced the line: >> >> merge_unicharsets $(TESSDATA)/$(CONTINUE_FROM).lstm-unicharset >> $(TRAIN)/my.unicharset "$@" >> >> with: >> >> cp "$(TRAIN)/my.unicharset" "data/unicharset" >> >> in the makefile (see >> https://groups.google.com/forum/#!searchin/tesseract-ocr/l.bolzani%7Csort:date/tesseract-ocr/be4-rjvY2tQ/32evtMHlAQAJ >> ) >> >> Then follow the instructions on the ocrd site. >> >> You can try 100, 250, 500, 1000 and 2000 iterations and see what works >> best (it depends on how much data you have). >> >> >> If you need to recognize nothing but handwritten numbers, you can also >> look for github projects (not related to tesseract) about "MNIST" >> handwritten numbers recognition with pre-trained models. >> >> >> Bye >> >> Lorenzo >> >> >> Il giorno lun 4 feb 2019 alle ore 08:34 <[email protected]> ha scritto: >> >>> I am a beginner for OCR training. Can anyone explain how to use Ocr-d >>> train briefly? >>> >>> I have Tesseract and Leptonica library installed in Cygwin >>> >>> tesseract 4.0.0 >>> leptonica-1.77.0 >>> libgif 5.1.4 : libjpeg 8d (libjpeg-turbo 1.5.3) : libpng 1.6.34 : >>> libtiff 4.0.9 : zlib 1.2.11 : libwebp 0.6.1 : libopenjp2 2.3.0 >>> Found AVX2 >>> Found AVX >>> Found SSE >>> >>> I want to train handwritten digits, because it is not detecting >>> correctly by default traineddata. I have searched group and found no >>> detailed instructions. I used Opencv and python tesseract combination to >>> achieve OCR of printed text and came to linux for handwritten digits >>> training purpose. Kindly provide step by step instructions, it may help >>> others also. I have attached the sample images which requires training. >>> Thanks in advance >>> >>> -- >>> You received this message because you are subscribed to the Google >>> Groups "tesseract-ocr" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to [email protected]. >>> To post to this group, send email to [email protected]. >>> Visit this group at https://groups.google.com/group/tesseract-ocr. >>> To view this discussion on the web visit >>> https://groups.google.com/d/msgid/tesseract-ocr/48ce49cc-6ade-4ebd-a1a6-5e382b033a95%40googlegroups.com >>> <https://groups.google.com/d/msgid/tesseract-ocr/48ce49cc-6ade-4ebd-a1a6-5e382b033a95%40googlegroups.com?utm_medium=email&utm_source=footer> >>> . >>> For more options, visit https://groups.google.com/d/optout. >>> >> -- > You received this message because you are subscribed to the Google Groups > "tesseract-ocr" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To post to this group, send email to [email protected]. > Visit this group at https://groups.google.com/group/tesseract-ocr. > To view this discussion on the web visit > https://groups.google.com/d/msgid/tesseract-ocr/4b4745ff-7bba-4982-8ced-6df1d03a4590%40googlegroups.com > <https://groups.google.com/d/msgid/tesseract-ocr/4b4745ff-7bba-4982-8ced-6df1d03a4590%40googlegroups.com?utm_medium=email&utm_source=footer> > . > For more options, visit https://groups.google.com/d/optout. > -- You received this message because you are subscribed to the Google Groups "tesseract-ocr" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/tesseract-ocr. To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/CABtjQ9KsD5qH%3D1peyf3dyx3PX023pqzrVR37yE-MefFThZUtXQ%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
