Usage of trained model in OpenAlpr failed

[Ognjen Gojkovic]

Hello,

I am trying to train the model for OpenALPR to detect Serbian license plates. There is already serbian font data in the original repository for training, but it is not meant to recognize latin letter characters like ŠĐĆČŽ which are important on Serbian license plates. 

I have gathered around 2300 pictures for the traning and the preparation for training has gone without any issues. But when i try to run the train.py, this is the console output i get:

Two-Letter Country Code to Train: rs

Processing: ./rs/input/leu.serbia.exp0.box

Executing: /usr/local/bin/tesseract -l eng ./rs/input/leu.serbia.exp0.tif leu.serbia.exp0 nobatch box.train.stderr

Failed loading language 'eng'

Tesseract couldn't load any languages!

Could not initialize tesseract.

mv: cannot stat './leu.serbia.exp0.tr': No such file or directory

mv: cannot stat './leu.serbia.exp0.txt': No such file or directory

Extracting unicharset from box file ./rs/input/leu.serbia.exp0.box

Other case a of A is not in unicharset

Other case b of B is not in unicharset

Other case c of C is not in unicharset

Other case d of D is not in unicharset

Other case e of E is not in unicharset

Other case f of F is not in unicharset

Other case g of G is not in unicharset

Other case h of H is not in unicharset

Other case i of I is not in unicharset

Other case j of J is not in unicharset

Other case k of K is not in unicharset

Other case l of L is not in unicharset

Other case m of M is not in unicharset

Other case n of N is not in unicharset

Other case o of O is not in unicharset

Other case p of P is not in unicharset

Other case r of R is not in unicharset

Other case s of S is not in unicharset

Other case t of T is not in unicharset

Other case u of U is not in unicharset

Other case v of V is not in unicharset

Other case w of W is not in unicharset

Other case x of X is not in unicharset

Other case y of Y is not in unicharset

Other case z of Z is not in unicharset

Other case ć of Ć is not in unicharset

Other case č of Č is not in unicharset

Other case đ of Đ is not in unicharset

Other case š of Š is not in unicharset

Other case ž of Ž is not in unicharset

Wrote unicharset file unicharset

Executing: /usr/bin/mftraining -F ./tmp/font_properties -U unicharset -O ./tmp/lrs.unicharset ./tmp/*.tr

Warning: No shape table file present: shapetable

Reading ./tmp/*.tr ...

Failed to open tr file: ./tmp/*.tr

Flat shape table summary: Number of shapes = 0 max unichars = 0 number with multiple unichars = 0

Warning: no protos/configs for Joined in CreateIntTemplates()

Warning: no protos/configs for |Broken|0|1 in CreateIntTemplates()

Warning: no protos/configs for 0 in CreateIntTemplates()

Warning: no protos/configs for 1 in CreateIntTemplates()

Warning: no protos/configs for 2 in CreateIntTemplates()

Warning: no protos/configs for 3 in CreateIntTemplates()

Warning: no protos/configs for 4 in CreateIntTemplates()

Warning: no protos/configs for 5 in CreateIntTemplates()

Warning: no protos/configs for 6 in CreateIntTemplates()

Warning: no protos/configs for 7 in CreateIntTemplates()

Warning: no protos/configs for 8 in CreateIntTemplates()

Warning: no protos/configs for 9 in CreateIntTemplates()

Warning: no protos/configs for A in CreateIntTemplates()

Warning: no protos/configs for B in CreateIntTemplates()

Warning: no protos/configs for C in CreateIntTemplates()

Warning: no protos/configs for D in CreateIntTemplates()

Warning: no protos/configs for E in CreateIntTemplates()

Warning: no protos/configs for F in CreateIntTemplates()

Warning: no protos/configs for G in CreateIntTemplates()

Warning: no protos/configs for H in CreateIntTemplates()

Warning: no protos/configs for I in CreateIntTemplates()

Warning: no protos/configs for J in CreateIntTemplates()

Warning: no protos/configs for K in CreateIntTemplates()

Warning: no protos/configs for L in CreateIntTemplates()

Warning: no protos/configs for M in CreateIntTemplates()

Warning: no protos/configs for N in CreateIntTemplates()

Warning: no protos/configs for O in CreateIntTemplates()

Warning: no protos/configs for P in CreateIntTemplates()

Warning: no protos/configs for R in CreateIntTemplates()

Warning: no protos/configs for S in CreateIntTemplates()

Warning: no protos/configs for T in CreateIntTemplates()

Warning: no protos/configs for U in CreateIntTemplates()

Warning: no protos/configs for V in CreateIntTemplates()

Warning: no protos/configs for W in CreateIntTemplates()

Warning: no protos/configs for X in CreateIntTemplates()

Warning: no protos/configs for Y in CreateIntTemplates()

Warning: no protos/configs for Z in CreateIntTemplates()

Warning: no protos/configs for Ć in CreateIntTemplates()

Warning: no protos/configs for Č in CreateIntTemplates()

Warning: no protos/configs for Đ in CreateIntTemplates()

Warning: no protos/configs for Š in CreateIntTemplates()

Warning: no protos/configs for Ž in CreateIntTemplates()

Done!

cp: cannot stat './rs/input/unicharambigs': No such file or directory

Reading ./tmp/*.tr ...

TrainingPage:Error:Assert failed:in file cntraining.cpp, line 125

Illegal instruction (core dumped)

mv: cannot stat './normproto': No such file or directory

Combining tessdata files

Output lrs.traineddata created successfully.

Version string:4.1.1

1:unicharset:size=2392, offset=192

3:inttemp:size=221590, offset=2584

4:pffmtable:size=285, offset=224174

13:shapetable:size=4, offset=224459

23:version:size=5, offset=224463

mv: cannot stat './lrs.normproto': No such file or directory

mv: cannot stat './lrs.unicharambigs': No such file or directory

After that, the lrs.traineddata file is generated, and when i try to import it and use it, i get this message from console:

tessdata_manager.SeekToStart(TESSDATA_NORMPROTO):Error:Assert failed:in file adaptmatch.cpp, line 561

fish: “python3 demo.py” terminated by signal SIGSEGV (Address boundary error)

I assume that something went wrong with the training because i followed all the steps exactly after gathering ~25 images of printed letters.


Also, i don't know if there will be a problem with detecting those special letters since the character ticks are above all other letters, and not under other letters as shown or .tif image. ('ve attatched some examples of how letters look on license plates)

Files for training are attached.

If you can train it for me, it would be much appreciated.

[Nam Phạm Quang]

Hi  Ognjen Gojkovic,

Did you train lrs.traineddata follow this tutorial?

https://github.com/openalpr/train-ocr

If true, please show me how did you set some parameters:

TESSERACT_DIR=''"
os.environ["TESSDATA_PREFIX"] = TESSERACT_DIR
TESSERACT_BIN=TESSERACT_DIR + '/tesseract'
TESSERACT_TRAINDIR= TESSERACT_DIR + '/training'

Many thanks.

Vào lúc 21:00:06 UTC+7 ngày Thứ Hai, 3 tháng 8, 2020, Ognjen Gojkovic đã viết:

[Ognjen Gojkovic]

Hello, thanks for your response and willingness to help,
Yes, i have trained it that way.

These were my parameters:

TESSERACT_DIR='/usr/share/tesseract-ocr/4.00/tessdata'

os.environ["TESSDATA_PREFIX"] = TESSERACT_DIR

TESSERACT_BIN= '/usr/local/bin/tesseract'

TESSERACT_TRAINDIR= '/usr/bin'

I have precompiled everything and tested, so it is working fine, but seems like there's some problem with training specifically.

Contents of TESSERACT_DIR:

[Nam Phạm Quang]

Hi Ognjen Gojkovic,

Thanks for your help.

I had run my trainning script, but still some problems, I will try to fix it.

Thanks again!

Vào lúc 23:11:48 UTC+7 ngày Thứ Năm, 24 tháng 9, 2020, Ognjen Gojkovic đã viết:

[Ognjen Gojkovic]

Thanks, let me know when you do something.

[Vladimir Radnovic]

Pozdrav Ognjene

postoji vec istrenirana bibljoteka, koja radi sa nasim slovima, zavisnosti za sta ti treba mozes se javiti na private msg pa cu ti detaljnije objasniti
pozdrav

[Ognjen Gojkovic]

Pozdrav Vladimire,
Hvala na interesovanju.
Iz nekog razloga ne mogu da vidim tvoju punu e-mail adresu, pa ako mozes, posalji mi poruku na ognjengo...@gmail.com, da tamo nastavimo.