项目作者: Swall0w

项目描述 :
crnn implemented in chainer
高级语言: Python
项目地址: git://github.com/Swall0w/chainer-crnn.git
创建时间: 2018-02-04T13:56:32Z
项目社区:https://github.com/Swall0w/chainer-crnn

开源协议:MIT License

下载


Convolutional Recurrent Neural Network

This repository implements the Convolutional Recurrent Neural Network(CRNN) in chainer.

You can find original implementation
and Pytorch implementation.

Demo

The demo script for CRNN can be found in demo.py.
Before running the demo, download a pretrained model from here.
This pretrained model is trained using train script.

The demo without any augments reads an example image and recognizes its text content.

Put the downloaded model file crnn.weights into directory data/.
Then launch the demo by:

  1. python3 demo.py
  2. a-----v--a-i-l-a-bb-l-ee-- => available

Prerequisites

  • Python 3.6.1 :: Anaconda custom (64-bit)
  • chainer >= 3.00
  • cupy
  • numpy

Train

I will write how to setup dataset and train model later.

Dataset for training is Synthetic Word Dataset

Author

  • Swall0w

License

  • MIT License