by deeperic

deeperic /SpikeFlow

Chinese OCR with TensorFlow

131 Stars 51 Forks Last release: Not found MIT License 10 Commits 0 Releases

Available items

No Items, yet!

The developer of this repository has not created any items for sale yet. Need a bug fixed? Help with integration? A different license? Create a request here:


A Chinese OCR with TensorFlow

*** Warning: The source codes in this repository may not work well with the latest version of Tensorflow. ***

To play around, follow these steps:

1/ Use Ocropy to generate Chinese character images. linegen is the tool used. You will need a font file. Put the images under folder training-character.gh.

2/ Run labelling-character.py to generate the labels on images.

3/ Run tf/convert-to-tfrecords.py to convert the images and labels in Tensorflow format.

4/ Modify the tf/helper.py for the characters you want to recognise.

5/ Run tf/train_model.py to train a model. The training will save a checkpoint on a regular interval.

6/ In receipt, run the findcontourcharacter.py to generate the images which may contain Chinese characters. You will have a "bw" folder containing all images. Run: python findcontourcharacter.py {image filename}

7/ Test the model by running: python testonechar.py {the name of your model} {the image to be recognised}

Blog: https://deeperic.wordpress.com/2017/02/18/chinese-ocr-tensorflow

Youtube: https://youtu.be/9N9OUruPZd4

GitHub: https://github.com/deeperic/SpikeFlow

We use cookies. If you continue to browse the site, you agree to the use of cookies. For more information on our use of cookies please see our Privacy Policy.