tesseract: Failed to read continue from: data/checkpoints/MODEL_NAME_checkpoint

Tesseract : tesseract 4.0.0 Leptonica: leptonica-1.77.0 Libs: 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

Platform: Windows 10 , cygwin 64bit

Hello, I was searching for my problem but could not find anything that matches.

While “make training”, using OCR-D I got this error :

Loaded file data/eng/eng.lstm, unpacking...
Warning: LSTMTrainer deserialized an LSTMRecognizer!
Code range changed from 111 to 111!
Num (Extended) outputs,weights in Series:
1,36,0,1:1, 0
Num (Extended) outputs,weights in Series:
C3,3:9, 0
Ft16:16, 160
Total weights = 160
[C3,3Ft16]:16, 160
 Mp3,3:16, 0
 Lfys64:64, 20736
 Lfx96:96, 61824
 Lrx96:96, 74112
 Lfx512:512, 1247232
Fc111:111, 0
Total weights = 1404064
Previous null char=110 mapped to 110
Continuing from data/eng/eng.lstm
Loaded 1/1 pages (1-1) of document data/ground-truth/tra1.Arial.exp2.lstmf
Loaded 1/1 pages (1-1) of document data/ground-truth/tra1.Arial.exp4.lstmf
Loaded 1/1 pages (1-1) of document data/ground-truth/tra1.Arial.exp1.lstmf
Loaded 1/1 pages (1-1) of document data/ground-truth/tra1.Arial.exp7.lstmf
Loaded 1/1 pages (1-1) of document data/ground-truth/tra1.Arial.exp6.lstmf
Loaded 1/1 pages (1-1) of document data/ground-truth/tra1.Arial.exp0.lstmf
!int_mode_:Error:Assert failed:in file /cygdrive/d/cyg_pub/devel/tesseract/tesseract-ocr-4.0.0- 
1.x86_64/src/tesseract-4.0.0/src/lstm/weightmatrix.cpp, line 249
!int_mode_:Error:Assert failed:in file /cygdrive/d/cyg_pub/devel/tesseract/tesseract-ocr-4.0.0- 
1.x86_64/src/tesseract-4.0.0/src/lstm/weightmatrix.cpp, line 249
!int_mode_:Error:Assert failed:in file /cygdrive/d/cyg_pub/devel/tesseract/tesseract-ocr-4.0.0- 
1.x86_64/src/tesseract-4.0.0/src/lstm/weightmatrix.cpp, line 249
 !int_mode_:Error:Assert failed:in file /cygdrive/d/cyg_pub/devel/tesseract/tesseract-ocr-4.0.0- 
1.x86_64/src/tesseract-4.0.0/src/lstm/weightmatrix.cpp, line 249
   lstmtraining \
  --stop_training \
  --continue_from data/checkpoints/tra1_checkpoint \
  --traineddata data/tra1/tra1.traineddata \
  --model_output data/tra1.traineddata
   **Failed to read continue from: data/checkpoints/tra1_checkpoint
  make: *** [Makefile:155: data/tra1.traineddata] Error** 1.

Its while executing this command:

lstmtraining \
  --traineddata $(PROTO_MODEL) \
      --old_traineddata $(TESSDATA)/$(START_MODEL).traineddata \
  --continue_from data/$(START_MODEL)/$(START_MODEL).lstm \
  --net_spec "[1,36,0,1 Ct3,3,16 Mp3,3 Lfys48 Lfx96 Lrx96 Lfx256 O1c`head -n1 data/unicharset`]" \
  --model_output data/checkpoints/$(MODEL_NAME) \
  --learning_rate 20e-4 \
  --train_listfile data/list.train \
  --eval_listfile data/list.eval \
  --max_iterations 5000

WHERE:

LOCAL := $(PWD)
PATH := $(LOCAL)/bin:$(PATH)

TESSDATA =  $(LOCAL)/data
MODEL_NAME = tra1
START_MODEL = eng

 LAST_CHECKPOINT = data/checkpoints/$(MODEL_NAME)_checkpoint
 PROTO_MODEL = data/$(MODEL_NAME)/$(MODEL_NAME).traineddata

I dont know if its neccessary but here is structure of dirs:

+---ocrd-->makefile and others, eng.traineddata -> **run "make training" from here
|   +---data        ----->all-lstmf/boxes,list.eval, list.train,unicharset
|   |   +---checkpoints
|   |   +---eng   -------->eng. traineddata eng.lstm->files, eng *-dawg->files
|   |   +---ground-truth-> my unicharset, tif , gt.txt , box , lstm files
|   |   +---lstmtraining
|   |   +---output
|   |   |   \---tra1
|   |   \---tra1 --> charset size, tra1.traineddata, tra1.unicharset
|   \---usr

I used chmod 777 and chown on checkpoints (as in Windows changed permissions) but it changed nothing. The eng.traineddata should be new aswell, but it doesnt look like to be a problem if it isnt. Also changed checkpoints dir location… Anyone can help with this one please? Fighting with this two days.

I will provide makefile code if neccesary.

About this issue

  • Original URL
  • State: closed
  • Created 8 years ago
  • Comments: 20

Most upvoted comments

Also try the following:

lstmtraining \
 --stop_training \
 --continue_from ~/tess/scratch/training0/output/jpn/scr_checkpoint \
 --traineddata ~/tess/scratch/training0/jpn/jpn.traineddata \
 --model_output $TESSDATA_PREFIX/scrjpninit.traineddata

No _ in the output traineddata (I remember having had issues when the output file name had _ in it).