lstmtraining command line related

[notorio...@gmail.com]

 I type the command line in my computer ubuntu 16.04.03 LTS

sudo lstmtraining --debug_interval -1 --traineddata /usr/share/tesseract-ocr/4.00/tessdata/kor.traineddata --net_spec '[1,0,0,1 Ct5,5,16 Mp3,3 Lfys64 Lfx128 Lrx128 Lfx256 01c105]' --train_listfile /usr/share/tesseract-ocr/4.00/tessdata/tesseract/training/trained_plus_chars_kor/kor.training_files.txt --eval_listfile /usr/share/tesseract-ocr/4.00/tessdata/tesseract/training/eval_plus_chars_kor/kor.training_files.txt --max_iterations 5000 

I have an error .

like 

Invalid network spec:01c105]

Missing ] at end of [Series]!

Failed to create network from spec: [1,0,0,1 Ct5,5,16 Mp3,3 Lfys64 Lfx128 Lrx128 Lfx256 01c105]

But, I saw the wiki page

https://github.com/tesseract-ocr/tesseract/wiki/VGSLSpecs

Full Example: A 1-D LSTM capable of high quality OCR

[1,1,0,48 Lbx256 O1c105]

As layer descriptions: (Input layer is at the bottom, output at the top.)

O1c105: Output layer produces 1-d (sequence) output, trained with CTC,
  outputting 105 classes.
Lbx256: Bi-directional LSTM in x with 256 outputs
1,1,0,48: Input is a batch of 1 image of height 48 pixels in greyscale, treated
  as a 1-dimensional sequence of vertical pixel strips.
[]: The network is always expressed as a series of layers.

This network works well for OCR, as long as the input image is carefully normalized in the vertical direction, with the baseline and meanline in constant places.

Full Example: A multi-layer LSTM capable of high quality OCR

[1,0,0,1 Ct5,5,16 Mp3,3 Lfys64 Lfx128 Lrx128 Lfx256 O1c105]

As layer descriptions: (Input layer is at the bottom, output at the top.)

O1c105: Output layer produces 1-d (sequence) output, trained with CTC,
  outputting 105 classes.
Lfx256: Forward-only LSTM in x with 256 outputs
Lrx128: Reverse-only LSTM in x with 128 outputs
Lfx128: Forward-only LSTM in x with 128 outputs
Lfys64: Dimension-summarizing LSTM, summarizing the y-dimension with 64 outputs

Mp3,3: 3 x 3 Maxpool
Ct5,5,16: 5 x 5 Convolution with 16 outputs and tanh non-linearity
1,0,0,1: Input is a batch of 1 image of variable size in greyscale
[]: The network is always expressed as a series of layers.

I have no idea .. why I type [ ] these charcter put in there . Take place an error 

Could you help me .?? 

[notorio...@gmail.com]

Invalid network spec:01c105]

Missing ] at end of [Series]!

Failed to create network from spec: [1,0,0,1 Ct5,5,16 Mp3,3 Lfys64 Lfx128 Lrx128 Lfx256 01c105]

2018년 3월 28일 수요일 오후 3시 53분 17초 UTC+9, notorio...@gmail.com 님의 말:

[shree]

PLEASE DO NOT SHOUT - Sending messages in Large fontsize, RED color etc is not appreciated. 

You have used a 0-zero instead of a CAPITAL O in your network spec, it should be O1c105

[이경준]

Okay .. ㅜㅜ Sorry I observed rule 

Thank You

--
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+unsubscribe@googlegroups.com.
To post to this group, send email to tesser...@googlegroups.com.
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/e1b97153-13b9-40d6-b583-417a13ace47e%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[notorio...@gmail.com]

@Shree

I want to make a traineddata

Could I have one more question about training from scratch ?

I execute that command line lstmtraining —debug_interval -1 —traineddata /usr/share/tesseract-ocr/4.00/tessdata/kor.traineddata —model_output /home/inplat/tesstutorial/koroutput/base —learning_rate 20e-4 —net_spec '1,0,0,1 Ct5,5,16 Mp3,3 Lfys64 Lfx128 Lrx128 Lfx256 O1c105' —train_listfile /usr/share/tesseract-ocr/4.00/tessdata/tesseract/training/trained_plus_chars_kor/kor.training_files.txt —eval_listfile /usr/share/tesseract-ocr/4.00/tessdata/tesseract/training/eval_plus_chars_kor/kor.training_files.txt —max_iterations 5000> /home/inplat/tesstutorial/koroutput/basetrain.log

And I have to do what steps for making traineddata .

I saw the wiki page . but I have no idea ㅠㅠ

2018년 3월 29일 목요일 오후 1시 59분 17초 UTC+9, 이경준 님의 말: