Simpleocr library
Simpleocr is a traditional chinese OCR python package that based on deep learning method.
The library consists of text localization and text recognition.
Text localization
The model is a reimplementation of CRAFT(Character-Region Awareness For Text detection) by tensorflow.
Text recognition
The reimplementation is based on CRNN model that RNN layer is replaced with self-attention layer.
CRNN
Self attention
Installation
$ pip install simpleocr
or
$ git clone https://github.com/xianyuntang/simpleocr
$ cd simpleocr
$ python setup.py install
Usage
from simpleocr import ocr
ocr.get_text(['image.jpg'])
TODO
- English support
- GPU support
