OCR program for Korean, SNU project 2016
- python 3
- flask
- numpy
- scipy
- tensorflow
- hangul-utils
- opencv-python
$ python app.py
- python 3.4
- numpy
- scipy
- tensorflow
- hangul-utils
python trainer.py
-i <tar data>
-o <path to save>
-e <epoch to train>
-b <mini batch size>
By default, 4 epochs and 100 per batch.
$ python trainer.py -i data/tar/161020.tgz -o data/ckpt/temp.ckpt