GenerateTrainingSet.ipynb · wang3960/zheye

加入 Gitee

与超过 1200万开发者一起发现、参与优秀开源项目，私有仓库也完全免费：）

该仓库未声明开源许可证文件（LICENSE），使用请关注具体项目描述及其代码上游依赖。

克隆/下载

GenerateTrainingSet.ipynb 44.02 KB

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 寻找合适的threshold"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 真实验证码"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 211,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 85,  99, 107, 117, 124, 126, 135, 140, 144, 151, 158, 162, 168,\n",
       "       173, 180, 186, 197, 203, 208, 215, 221, 225, 232, 249], dtype=uint8)"
      ]
     },
     "execution_count": 211,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "im = Image.open('realcap/captcha (10).gif').convert('L')\n",
    "X_im = np.asarray(im).copy()\n",
    "uniques, counts = np.unique(X_im, return_counts=True)\n",
    "uniques"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 212,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiYAAAFkCAYAAAAUtvC8AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYTCRJEnFMJhIkqRi/H9bqEp+1HM+LwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f55d77d0550>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from pylab import *\n",
    "scatter(uniques, counts)\n",
    "show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 213,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "X_im[X_im <= 150] = 0\n",
    "X_im[X_im > 150] = 255"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 214,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZAAAABYCAAAAAD+GP59AAAHlElEQVR4nO1dybbjKAxFfer/f1m9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\nYII=\n",
      "text/plain": [
       "<PIL.Image.Image image mode=L size=400x88 at 0x7F55D7831710>"
      ]
     },
     "execution_count": 214,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image.fromarray(X_im)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 模拟生成验证码"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 274,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "r = util.RandomGenerateOneChar(y=1).convert('L')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 275,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAACgAAAAoCAAAAACpleexAAADt0lEQVR4nG2UW4jVVRTGf3v/b2eO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\neqPC5tQvgbJF7QIwBvBfIwD+B4AHYkqY+QnhAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<PIL.Image.Image image mode=L size=40x40 at 0x7F55D7B5F050>"
      ]
     },
     "execution_count": 275,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 276,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X = np.asarray(r).astype('float64')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 277,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "uniques, counts = np.unique(X, return_counts=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 278,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAh4AAAFkCAYAAABvkjJwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMi8AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f55d7d60fd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from pylab import *\n",
    "scatter(uniques, counts)\n",
    "show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 279,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X[X <= 150] = 0\n",
    "X[X > 150] = 255"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 242,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAACgAAAAoCAAAAACpleexAAAA20lEQVR4nK2USxaAIAwDUx73v3Jc\nKNCUVlnIRgtD+lUjYABoNBC3IYsGEOiPabAdontvwfbY3CcAo9dxRry9KBrnBgNDJAom2Dhn6Urd\nAV0CKzGwFylrilh1rKSmjoCxVn41h5meauTm+8sAxRbmvoLdDrmVTFWmEUlWnshMxb0xd/6y3zMO\nwlGUs5GhH7keT79i3YdoduKWzL4ZZqVqjmONAe290g703Nud9nL2F1j8V/53/TlmI5pckatU43nk\nWn+wYQgepxzOVXFrzQrAakhXi0GX4PH0nGHABSQsSEGRYpu+AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<PIL.Image.Image image mode=L size=40x40 at 0x7F55D7D36050>"
      ]
     },
     "execution_count": 242,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image.fromarray(X)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 生成训练数据"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from PIL import Image, ImageFont, ImageDraw\n",
    "from zheye import util\n",
    "from random import randint\n",
    "import numpy as np\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sample_size = 1000"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "X = []\n",
    "Y = []\n",
    "\n",
    "for i in range(sample_size):\n",
    "    #direction = np.random.binomial(1, 0.5)\n",
    "    #if direction == 0:\n",
    "    #    direction = -1\n",
    "    direction = -1\n",
    "    \n",
    "    ret = util.RandomGenerateOneChar(y=direction)\n",
    "\n",
    "    tmp = np.asarray(ret.convert('L')).astype('float64')\n",
    "    \n",
    "    tmp[tmp <= 150] = -1\n",
    "    # 黑色 \n",
    "    tmp[tmp >  150] = 1\n",
    "    # 白底\n",
    "    \n",
    "    X.append(tmp)\n",
    "    \n",
    "    if direction == 1:\n",
    "        Y.append([1,0])\n",
    "    else:\n",
    "        Y.append([0,1])\n",
    "\n",
    "X = np.array(X).astype(\"float64\")\n",
    "Y = np.array(Y).astype(\"uint8\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0 1]\n"
     ]
    }
   ],
   "source": [
    "print Y[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1000, 2)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Y.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 标准化是依照特征矩阵的列处理数据(to do)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 1.,  1.,  1., ..., -1.,  1.,  1.],\n",
       "       [-1.,  1.,  1., ...,  1.,  1.,  1.],\n",
       "       [-1., -1., -1., ...,  1.,  1.,  1.],\n",
       "       ..., \n",
       "       [ 1.,  1.,  1., ...,  1.,  1.,  1.],\n",
       "       [ 1.,  1.,  1., ...,  1.,  1.,  1.],\n",
       "       [ 1.,  1.,  1., ...,  1.,  1.,  1.]])"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "''"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\"\"\"\n",
    "mean = (X.sum(axis=0) /sample_size)\n",
    "Image.fromarray(mean)\n",
    "np.sum(mean)/mean.size\n",
    "X = (X.astype(\"float64\") - mean)/256\n",
    "\"\"\"\n",
    "\"\"\"\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1000, 40, 40, 1)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = np.expand_dims(X, axis=3)\n",
    "X.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 存储"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 302,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "np.save('test_X', X)\n",
    "np.save('test_Y', Y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "np.save('train_X', X)\n",
    "np.save('train_Y', Y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 生成召回率/查准率样本\n",
    "\n",
    "$$\n",
    "\\begin{array}{c|lll}\n",
    "n & \\text{正样本} & \\text{负样本} \\\\\n",
    "\\hline\n",
    "\\text{识别为正样本} & \\text{True Positive} & \\text{False Positive(一类错误)} \\\\\n",
    "\\text{识别为负样本} & \\text{False Negative(二类错误)} & \\text{True Negative}\n",
    "\\end{array}|\n",
    "$$\n",
    "\n",
    "查准率$ Precision = TP/(TP+FP) $\n",
    "\n",
    "召回率$ Recall = TP/(TP+FN) $"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "np.save('testF_X', X)\n",
    "np.save('testF_Y', Y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "np.save('testT_X', X)\n",
    "np.save('testT_Y', Y)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}

一键复制编辑原始数据按行查看历史

提交于 2017-05-05 15:18 . compute Recall

马建仓 AI 助手

尝试更多

代码解读

代码找茬

代码优化

https://gitee.com/wang3960/zheye.git

git@gitee.com:wang3960/zheye.git

wang3960

zheye

master

wang3960/zheye

简介

发行版

贡献者

近期动态

wang3960/zheye .gitee-modal { width: 500px !important; }

简介

发行版

贡献者

近期动态

搜索帮助

wang3960/zheye