work with Arabic language about synthtiger HOT 2 CLOSED

First you need to prepare Arabic corpus file and font files.
Put text file (.txt) in resources/corpus path and font files (.ttf) in resources/font path.

Next, run font preprocessing script.

python tools/extract_font_charset.py -w 4 resources/font

And then, you need to edit config file.
Change corpus file path to prepared corpus file path.
https://github.com/clovaai/synthtiger/blob/master/examples/synthtiger/config_horizontal.yaml#L14-L15

- paths: [resources/corpus/arabic.txt]
  weights: [1]

You can change other parameter values (e.g. min_length: null, max_length: null, augmentation: 0).

Finally, run this command to generate the Arabic dataset.

synthtiger -o results -w 4 -v examples/synthtiger/template.py SynthTiger examples/synthtiger/config_horizontal.yaml

NOTE:

Arial font is good for generating Arabic text images. But, other Arabic fonts may not be supported.
This problem is related to SynthTiger.

If you want to use other Arabic fonts now, you should change code as below (172, 196 line in template.py).
Note that if you change this code, you can't render curved text.

# chars = utils.split_text(label, reorder=True)
chars = [label]

And then, run this command to generate the Arabic dataset.

python -m synthtiger -o results -w 4 -v examples/synthtiger/template.py SynthTiger examples/synthtiger/config_horizontal.yaml

from synthtiger.

thanks a lot for your fast replaying and your instructions

from synthtiger.