Repository files navigation

Generate text images for training deep learning OCR model (e.g.CRNN).Support both latin and non-latin text.

• Ubuntu 16.04
• python 3.5+

Install dependencies:

pip3 install -r requirements.txt

By default, simply runpython3 main.py will generate 20 text imagesand a labels.txt file inoutput/default/.

1. Please runpython3 main.py --help to see all optional arguments and their meanings.And put your own data in corresponding folder.

2. Config text effects and fraction inconfigs/default.yaml file(or create anew config file and use it by--config_file option), here are some examples:

Effect name	Image
Origin(Font size 25)
Perspective Transform
Random Crop
Curve
Light border
Dark border
Random char space big
Random char space small
Middle line
Table line
Under line
Emboss
Reverse color
Blur
Text color
Line color

3. Runmain.py file.

For no-latin language(e.g Chinese), it's very common that some fonts only supportlimited chars. In this case, you will get bad results like these:

Select fonts that support all chars in--chars_file is annoying.Runmain.py with--strict option, renderer will retry get text fromcorpus during generate processing until all chars are supported by a font.

You can usecheck_font.py script to check how many chars your font not support in--chars_file:

python3 tools/check_font.pychecking font ./data/fonts/eng/Hack-Regular.ttfchars not supported(4971):['第','朱','广','沪','联','自','治','县','驼','身','进','行','纳','税','防','火','墙','掏','心','内','容','万','警','钟','上','了','解'...]0 fonts support all chars(5071)in ./data/chars/chn.txt:[]

If you want to use GPU to make generate image faster, first compile opencv with CUDA.Compiling OpenCV with CUDA support

Then build Cython part, and add--gpu option when runmain.py

cd libs/gpupython3 setup.py build_ext --inplace

Runpython3 main.py --debug will save images with extract information.You can see how perspectiveTransform works and all bounding/rotated boxes.

Seehttps://github.com/Sanster/text_renderer/projects/1

If you use text_renderer in your research, please consider use the following BibTeX entry.

@misc{text_renderer,author ={weiqing.chu},title ={text_renderer},howpublished ={\url{https://github.com/Sanster/text_renderer}},year ={2021}}