Maybe it is just a bug I need to open an issue?
On Monday, September 20, 2021 at 2:52:18 PM UTC+3 Sim Tov wrote:
> Hello,
>
> I use v4.1.1 on Linux (Debian 11) and try to generate train and evaluate
> data. The commands I used were:
>
> train:
>
> usr/share/tesseract-ocr/tesstrain.sh --fonts_dir FontsRashi/Working --lang
> heb --linedata_only --noextract_font_properties --langdata_dir ./langdata
> --tessdata_dir /usr/share/tesseract-ocr/4.00/tessdata/ --output_dir
> output/train --fontlist 'BenOr Rashi' 'Guttman Rashi Bold'
>
> and
>
> evaluate:
>
> /usr/share/tesseract-ocr/tesstrain.sh --fonts_dir FontsRashi/Working
> --lang heb --linedata_only --noextract_font_properties --langdata_dir
> ./langdata --tessdata_dir /usr/share/tesseract-ocr/4.00/tessdata/
> --output_dir output/evaluate --fontlist 'Guttman Rashi'
>
> After several days of running both commands stopped with errors like this
> for each of the 3 fonts:
>
> Page 8365
> Loaded 386170/386170 lines (1-386170) of document
> /tmp/heb-2021-09-16.1dB/heb.Guttman_Rashi.exp0.lstmf
> Page 8366
> Loaded 386216/386216 lines (1-386216) of document
> /tmp/heb-2021-09-16.1dB/heb.Guttman_Rashi.exp0.lstmf
> /usr/share/tesseract-ocr/tesstrain_utils.sh: line 72: 2271 Segmentation
> fault "${cmd}" "$@" 2>&1
> 2272 Done | tee -a ${LOG_FILE}
> ERROR: Program tesseract failed. Abort.
>
> Interestingly that heb.Guttman_Rashi.exp0.lstmf and both others .lstmf
> files were exactly 1Gb big...
>
> Does it has something to do with what is written here:
>
>
> https://github.com/tesseract-ocr/tesseract/issues/654#issuecomment-274574951
>
> "The text is divided by language automatically, so there is a separate
> stream for each of the Devanagari-based languages (as there is for the
> Latin-based languages) and *clipped to 1GB *for each language."
>
> 1. So is this Segmentation fault an expected behavior?
>
> 2. What should I do now? Should I rerun the commands hoping that they will
> finish properly or should I copy those .lstmf files that I got so far to
> the train/evaluate directories and start training?
>
> 3. Both output/evaluate and output/train directories remained empty after
> the commands above failed. What files should be there at the end so I can
> start the training process?
>
>
> Thank you in advance!
>
> tesseract --version
> tesseract 4.1.1
> leptonica-1.79.0
> libgif 5.1.9 : libjpeg 6b (libjpeg-turbo 2.0.6) : libpng 1.6.37 :
> libtiff 4.2.0 : zlib 1.2.11 : libwebp 0.6.1 : libopenjp2 2.4.0
> Found AVX
> Found FMA
> Found SSE
> Found libarchive 3.4.3 zlib/1.2.11 liblzma/5.2.5 bz2lib/1.0.8
> liblz4/1.9.3 libzstd/1.4.8
>
>
--
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 on the web visit
https://groups.google.com/d/msgid/tesseract-ocr/cec596a9-cdfb-4a68-ab49-d275f27a82a5n%40googlegroups.com.
