Adding New Fonts to Tesseract 3

5
votes

I'm trying to add new fonts to tesseract ocr. I'm following this tutorial but I'm having some problems.

Here's what I've done so far:

  1. Create training document

    convert eng.myfont.exp0.pdf eng.myfont.exp0.tif

  2. Train Tesseract

    tesseract eng.myfont.exp0.tif eng.myfont.exp0 batch.nochop makebox

    This created my eng.myfont.exp0.box file.

    I open the file with moshpytt and make sure it was detected correctly.

  3. Feed the box file back into tesseract

    tesseract eng.myfont.exp0.tif eng.myfont.exp0.box nobatch box.train.stderr

    I have this result:

    Tesseract Open Source OCR Engine v3.03 with Leptonica
    APPLY_BOXES:
    Boxes read from boxfile: 146
    Found 146 good blobs.
    TRAINING ... Font name = myfont.exp0
    Generated training data for 6 words

    • eng.myfont.exp0.box.tr file and eng.myfont.exp0.box.txt generated

  4. try to detect the Character set used in the box file (this is where I get stuck)

    unicharset_extractor *.box

Result:

unicharset_extractor: command not found

I also tred unicharset_extractor eng.myfont.exp0.box with the same result.

I'm using:

  • tesseract 3.03
  • leptonica-1.70
  • libgif 4.1.6(?) : libjpeg 8d : libpng 1.2.50 : libtiff 4.0.3 : zlib 1.2.8 : webp 0.4.0
  • Ubuntu 14.04.1 LTS
2
ocrtesseract
That's pretty peculiar. It just means the command cannot be found. On my system I'm able to find that command without any issue in /usr/local/bin/unicharset_extractor. – mlissner

2 Answers

5
votes

The training tools for Tesseract 3.03 RC were omitted from Ubuntu 14.04. So either fall back to Tesseract 3.02 or upgrade to Ubuntu 14.10, which should have it.

3
votes

Ok, I googled this for you. Here's the answer:

You need to run all commands in the same folder where are located your input files.

From: