Bazaar & eng.user-patterns not doing anything?

[phillip...@gmail.com]

Hi,

I created a bazaar file as attached with load_system_dawg, load_freq_dawg set to F. I also want to use user-patterns so I set it as well.

load_system_dawg F
load_freq_dawg F
user_patterns_suffix user-patterns

In the same directory, I also have the user-pattern file:

\d\d\d\d\c

So the structure looks like:

./bazaar

./eng.user-patterns

./ocr_inv.png

 

But these settings still fail to recognise the image correctly as "1880A". If I just run tesseract without any bells and whistles, the outputs are still the same.

Commands used:

tesseract ocr_inv.png stdout
tesseract ocr_inv.png stdout bazaar
tesseract ocr_inv.png stdout --user-patterns eng.user-patterns bazaar

Output:

Warning. Invalid resolution 0 dpi. Using 70 instead.

Estimating resolution as 1128

ISON

Can anyone tell me if this is expected behaviour?

Tesseract version:

tesseract 4.0.0-beta.1

 leptonica-1.75.3

  libgif 5.1.4 : libjpeg 8d (libjpeg-turbo 1.5.2) : 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

[Zdenko Podobny]

First of all: this is very old tesseract version. Try recent code, where were a lot of bug fixes (not sure if for bazar/user patterns  too. 

Zdenko

pi 26. 4. 2019 o 11:53 <phillip...@gmail.com> napísal(a):

--
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 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/10d1ed21-d30e-4eaf-8f2a-6fdf74a6a7d1%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[JB Data31]

Image processing can improve the result, but this typo is very particular, i.e. "unconnected" digit.

I try morphological transformations to re-connect digit, better (tesseract ocr_inv-8.png ocr-8 --psm 6) , but yet far away a proper result.

According to me, train tesseract with this typo is a way.

@JBΔ

[phillip...@gmail.com]

I tried this version:

tesseract 4.0.0

 leptonica-1.75.3

  libgif 5.1.4 : libjpeg 8d (libjpeg-turbo 1.5.2) : 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

It still gives me the same outputs with or without the user-pattern and config files.

To unsubscribe from this group and stop receiving emails from it, send an email to tesser...@googlegroups.com.

[phillip...@gmail.com]

Did you have any luck without training, but including the bazaar config file and user-patterns txt file?

To unsubscribe from this group and stop receiving emails from it, send an email to tesser...@googlegroups.com.

[phillip...@gmail.com]

UPDATES:

I trained a file and put it in tessdata. The output is better now, but it seems that the user-pattern file is still not working?

Command:

tesseract ocr_inv.png ocr_inv -l Karton --user-patterns Karton.user-patterns Karfig

[phillip...@gmail.com]

UPDATES:

It works fine now. All I did is git clone the latest master branch and build from source. Then the user-pattern and configs can work.