Thanks for the reply. Yes, I also use jeTesBoxEditor at the same time, but jeTesBoxEditor is more like data standardization. Some of the font files have incomplete fonts. I want to use LSTM training to train a complete autologous library file of my own. 在2024年12月6日星期五 UTC+8 15:15:40<mahmoud...@gmail.com> 写道：
> I think using jeTesBoxEditor is good for training process > > في الجمعة، ٦ ديسمبر ٢٠٢٤، ١١:٠٧ ص Zdenko Podobny <zde...@gmail.com> كتب: > >> >> Error: Tesseract (legacy) engine requested, but components are not >> present in /usr/local/share/tessdata/my_chi_sim.traineddata!! >> >> >> The message is clear. YOU require tesseract to use legacy engine >> explicitly but YOUR language datafile (you created by training) does not >> contain legacy model. >> >> Zdenko >> >> >> pi 6. 12. 2024 o 7:11 鹿青年 <luqingn...@gmail.com> napísal(a): >> >>> Hello, I tried to train a traineddata file myself, but an [Error] >>> occurred during use. Could you please give me some guidance on how to >>> resolve this error? Thank you very much. >>> Perform OCR >>> ··· >>> tesseract 0791.tif stdout -l my_chi_sim --psm 6 --oem 2 >>> ··· >>> The error content is: >>> ···· >>> Error: Tesseract (legacy) engine requested, but components are not >>> present in /usr/local/share/tessdata/my_chi_sim.traineddata!! >>> Failed loading language 'my_chi_sim' >>> Tesseract couldn't load any languages! >>> Could not initialize tesseract. >>> ···· >>> >>> My training steps are as follows: >>> >>> Punctuation Dictionary: >>> dawg2wordlist d:\tesseract\tessdata_best\chi_sim.lstm-unicharset >>> d:\tesseract\tessdata_best\chi_sim.lstm-punc-dawg >>> d:\tesseract\tessdata_best\punc.txt >>> >>> >>> Let’s start with the key steps >>> 2. Generate character set lstm-unicharset file >>> 1. Generate character set txt file >>> >>> text2image --text d:\tesseract\chi_sim.txt --outputbase >>> d:\tesseract\chi_sim --fonts_dir C:\Windows\Fonts --font="simhei" >>> --fontconfig_tmpdir d:\tesseract\tmp >>> >>> >>> 3. Generate character set lstm-unicharset file >>> >>> 1) Generate with box file >>> unicharset_extractor --norm_mode 3 --output_unicharset >>> d:\tesseract\chi_sim.lstm-unicharset d:\tesseract\chi_sim.box >>> >>> 2) Generate with txt file >>> unicharset_extractor --norm_mode 3 --output_unicharset >>> d:\tesseract\chi_sim.lstm-unicharset d:\tesseract\chi_sim.txt >>> >>> >>> 3. Generate starter traineddata file >>> 1. Generate dictionary text file >>> Refer to the 3 dictionary files in the d:\tesseract\tessdata_best folder >>> (word text, number numbers, punc punctuation marks) >>> 2. Generate starter traineddata file >>> combine_lang_model --input_unicharset >>> d:\tesseract\chi_sim.lstm-unicharset --lang chi_sim --script_dir >>> d:\tesseract\langdata_lstm --output_dir d:\tesseract --version_str >>> "CSDN:watt:2022.04[1,48,0,1C3,3Ft16Mp3,3TxyLfys64Lfx96RxLrx96Lfx512O1c4000]" >>> >>> --words d:\tesseract\word.txt --numbers d:\tesseract\number.txt --puncs >>> d:\tesseract\punc.txt --pass_through_recoder >>> >>> >>> 3. View the newly generated starter trained data information >>> combine_tessdata -d d:\tesseract\chi_sim\chi_sim.traineddata >>> >>> 4. Generate training files >>> 1. Generate the training text file train.txt >>> >>> 2. Generate picture+box file >>> >>> text2image --text d:\tesseract\train.txt --outputbase d:\tesseract\train >>> --fonts_dir C:\Windows\Fonts --font="simhei" --ptsize 18 >>> --fontconfig_tmpdir d:\tesseract\tmp >>> 3. Generate training files: >>> tesseract d:\tesseract\train.tif d:\tesseract\train -l chi_sim --psm 6 >>> lstm.train >>> >>> 4. Create a new training list file >>> Create a new d:\tesseract\train_listfile.txt file with the content >>> d:\tesseract\train.lstmf >>> 5. Training >>> >>> 2. Start training: >>> lstmtraining --traineddata d:\tesseract\chi_sim\chi_sim.traineddata >>> --net_spec "[1,48,0,1Ct3,3,16 Mp3,3 Lfys64 Lfx96 Lrx96 Lfx512 O1c4000]" >>> --model_output d:\tesseract\output\output --train_listfile >>> d:\tesseract\train_listfile.txt --max_iterations 0 --target_error_rate 0.01 >>> --debug_interval -1 >>> >>> 6. Evaluate the generated checkpoint file >>> 1. Generate evaluation text eval.txt >>> Edit some evaluation text and save it to d:\tesseract\eval.txt, so as to >>> cover it as comprehensively as possible and with a certain degree of >>> complexity. >>> 2. Generate picture+box file >>> text2image --text d:\tesseract\eval.txt --outputbase d:\tesseract\eval >>> --fonts_dir C:\Windows\Fonts --font="simhei" --ptsize 18 >>> --fontconfig_tmpdir d:\tesseract\tmp >>> 3. Generate evaluation lstmf file >>> tesseract d:\tesseract\eval.tif d:\tesseract\eval -l chi_sim --psm 6 >>> lstm.train >>> 4. Generate evaluation list file >>> Create a new d:\tesseract\eval_listfile.txt file with the content >>> d:\tesseract\eval.lstmf >>> 5. Start evaluating >>> >>> Start evaluating: >>> lstmeval --model d:\tesseract\output\output_checkpoint --traineddata >>> d:\tesseract\chi_sim\chi_sim.traineddata --eval_listfile >>> d:\tesseract\eval_listfile.txt >>> 7. Generate standard trained data >>> 1. Generate a floating point (decimal) traineddata file (similar to >>> tessdata_best) >>> lstmtraining --stop_training --continue_from >>> d:\tesseract\output\output_checkpoint --traineddata >>> d:\tesseract\chi_sim\chi_sim.traineddata --model_output >>> d:\tesseract\output\chi_sim.traineddata >>> 2. Generate an integer traineddata file (similar to tessdata_fast) >>> lstmtraining --stop_training --convert_to_int --continue_from >>> d:\tesseract\output\output_checkpoint --traineddata >>> d:\tesseract\chi_sim\chi_sim.traineddata --model_output >>> d:\tesseract\output\chi_sim.traineddata >>> >>> 3. View the generated traineddata information >>> combine_tessdata -d d:\tesseract\output\chi_sim.traineddata >>> >>> -- >>> 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 tesseract-oc...@googlegroups.com. >>> To view this discussion visit >>> https://groups.google.com/d/msgid/tesseract-ocr/4f54b4ff-f1f4-4e44-9e49-11a70b759d68n%40googlegroups.com >>> >>> <https://groups.google.com/d/msgid/tesseract-ocr/4f54b4ff-f1f4-4e44-9e49-11a70b759d68n%40googlegroups.com?utm_medium=email&utm_source=footer> >>> . >>> >> -- >> 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 tesseract-oc...@googlegroups.com. >> > To view this discussion visit >> https://groups.google.com/d/msgid/tesseract-ocr/CAJbzG8xMU7KTYPbKPTby09M6cEOvBbngD-U8hRyR%2BWZPF3_HhQ%40mail.gmail.com >> >> <https://groups.google.com/d/msgid/tesseract-ocr/CAJbzG8xMU7KTYPbKPTby09M6cEOvBbngD-U8hRyR%2BWZPF3_HhQ%40mail.gmail.com?utm_medium=email&utm_source=footer> >> . >> > -- 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 tesseract-ocr+unsubscr...@googlegroups.com. To view this discussion visit https://groups.google.com/d/msgid/tesseract-ocr/d7c9c21b-cc85-49c7-b7e3-2e40aad4cc35n%40googlegroups.com.
