# pvc

**Repository Path**: jtyoui/pvc

## Basic Information

- **Project Name**: pvc
- **Description**: 这个一个基于TensorFlow做的一个验证码识别器
- **Primary Language**: Python
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2018-01-25
- **Last Updated**: 2021-01-29

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# **pvc** [![tyoui](https://github.com/zhangwei0530/logo/blob/master/logo/photolog.png?raw=true)](http://www.tyoui.cn)

[![](https://github.com/zhangwei0530/logo/blob/master/logo/logo.png?raw=true)](http://www.tyoui.cn)

## PVC是一个基于TensorFlow做的一个验证码识别器

[![](https://img.shields.io/badge/language-python-orange.svg)]()
[![](https://img.shields.io/badge/Python-3.6-green.svg)]()
[![](https://img.shields.io/badge/BlogWeb-Tyoui-bule.svg)](http://www.tyoui.cn)
[![](https://img.shields.io/badge/Email-tyoui@tyoui.cn-red.svg)]()

## 使用的卷积神经网络流程图

![](https://github.com/zhangwei0530/logo/blob/master/photo/cnn.png?raw=true)

## 卷积神经网络代码
```python
def crack_captcha_cnn(w_alpha=0.01, b_alpha=0.1):
    x = tf.reshape(X, shape=[-1, IMAGE_HEIGHT, IMAGE_WIDTH, 1])
    w_c1 = tf.Variable(w_alpha * tf.random_normal([3, 3, 1, 32]))
    b_c1 = tf.Variable(b_alpha * tf.random_normal([32]))
    rel_1 = tf.nn.relu(tf.nn.bias_add(tf.nn.conv2d(x, w_c1, strides=[1, 1, 1, 1], padding='SAME'), b_c1))
    pool_1 = tf.nn.max_pool(rel_1, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
    con_1 = tf.nn.dropout(pool_1, keep_prob)
    w_c2 = tf.Variable(w_alpha * tf.random_normal([3, 3, 32, 64]))
    b_c2 = tf.Variable(b_alpha * tf.random_normal([64]))
    rel_2 = tf.nn.relu(tf.nn.bias_add(tf.nn.conv2d(con_1, w_c2, strides=[1, 1, 1, 1], padding='SAME'), b_c2))
    pool_2 = tf.nn.max_pool(rel_2, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
    con_2 = tf.nn.dropout(pool_2, keep_prob)
    w_c3 = tf.Variable(w_alpha * tf.random_normal([3, 3, 64, 64]))
    b_c3 = tf.Variable(b_alpha * tf.random_normal([64]))
    rel_3 = tf.nn.relu(tf.nn.bias_add(tf.nn.conv2d(con_2, w_c3, strides=[1, 1, 1, 1], padding='SAME'), b_c3))
    pool_3 = tf.nn.max_pool(rel_3, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
    con_3 = tf.nn.dropout(pool_3, keep_prob)
    w_d = tf.Variable(w_alpha * tf.random_normal([8 * 32 * 40, 1024]))
    b_d = tf.Variable(b_alpha * tf.random_normal([1024]))
    dense = tf.reshape(con_3, [-1, w_d.get_shape().as_list()[0]])
    sigmoid = tf.nn.sigmoid(tf.add(tf.matmul(dense, w_d), b_d))
    dense = tf.nn.dropout(sigmoid, keep_prob)
    w_out = tf.Variable(w_alpha * tf.random_normal([1024, MAX_CAPTCHA * CHAR_SET_LEN]))
    b_out = tf.Variable(b_alpha * tf.random_normal([MAX_CAPTCHA * CHAR_SET_LEN]))
    out = tf.add(tf.matmul(dense, w_out), b_out)
    return out
```
## 先执行train.py，之后会有训练模型model文件夹，输出是每训练10次打印一次精度
    0 精度: 0.0875
    10 精度: 0.11
  
当打印到百分之90时，将自动停止。训练精度设置在

```python
    if acc > 0.900:
        saver.save(sess, "./model/captcha.model", global_step=0)
        break
```
 
## 测试文件。在test.py
    测试时将自动生成一张照片。输出正确值与测试值
    
## 验证码文件值
    char_set = number + alphabet + ALPHABET