Introduction

ZFScripting is a scripting build by Python for logging in Zhengfang system.

Dependencies

  • request

Usage

1. Label the images

run the InputLabel.py and then open http://127.0.0.1:5000 in your Browser. The label you input will store in label.db and images in ./images/

2. Convert them to train data

run the DataTool.py, the images will be filter for noice reduction and each of them will be divided into 4 parts. Then each of part is reshaped into 1-D numpy array. Finally put them in X that is a 2-D numpy array. The labels you input which stored in label.db will be divided into 4 parts as well and mapped to integers range from 0 to 35, stored in Y, a 1-D numpy array.

3. Train the model

Versions in the future

Version things to do
0.1 get picture samples from ZF
0.2 reduce the noise of the pictures with image processing
0.4 obtain the labels from user input or download dataset from Internet
0.7 train a model from the dataset with probability over 50%
0.9 log in ZF with the CAPTCHA recognition program
1.0 finish the GUI section