main.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Enabling notebook extension jupyter-js-widgets/extension...\n",
      "      - Validating: ok\n"
     ]
    }
   ],
   "source": [
    "!jupyter nbextension enable --py widgetsnbextension"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found submodule _build (is a package: False)\n",
      "Found submodule _shared (is a package: True)\n",
      "Found submodule color (is a package: True)\n",
      "Found submodule data (is a package: True)\n",
      "Found submodule draw (is a package: True)\n",
      "Found submodule exposure (is a package: True)\n",
      "Found submodule external (is a package: True)\n",
      "Found submodule feature (is a package: True)\n",
      "Found submodule filters (is a package: True)\n",
      "Found submodule future (is a package: True)\n",
      "Found submodule graph (is a package: True)\n",
      "Found submodule io (is a package: True)\n",
      "Found submodule measure (is a package: True)\n",
      "Found submodule morphology (is a package: True)\n",
      "Found submodule novice (is a package: True)\n",
      "Found submodule restoration (is a package: True)\n",
      "Found submodule run-hessian (is a package: False)\n",
      "Found submodule scripts (is a package: True)\n",
      "Found submodule segmentation (is a package: True)\n",
      "Found submodule setup (is a package: False)\n",
      "Found submodule transform (is a package: True)\n",
      "Found submodule util (is a package: True)\n",
      "Found submodule viewer (is a package: True)\n"
     ]
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import pkgutil\n",
    "from sklearn.datasets import fetch_lfw_people\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.image as mpimg\n",
    "import numpy as np\n",
    "import skimage\n",
    "import skimage.io\n",
    "from skimage.feature import hog\n",
    "from sklearn.feature_extraction import image\n",
    "from tqdm import tqdm, tqdm_notebook, tnrange\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.svm import LinearSVC\n",
    "from sklearn.metrics import f1_score\n",
    "from sklearn.externals import joblib\n",
    "test = 123\n",
    "test\n",
    "\n",
    "\n",
    "package = skimage\n",
    "for importer, modname, ispkg in pkgutil.iter_modules(package.__path__): \n",
    "    print(\"Found submodule %s (is a package: %s)\" % (modname, ispkg))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "people = fetch_lfw_people(color=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(13233, 62, 47, 3)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "people.keys()\n",
    "\n",
    "people.images.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "filepath = \"C:\\\\Users\\\\danai\\\\Projects\\\\python projects\\\\PycharmProjects\\\\IT_1_coursework\\\\neg_ex\\\\космос.png\"\n",
    "file_ = skimage.io.imread(filepath)\n",
    "patches = image.extract_patches_2d(file_, people.images.shape[1:3], max_patches = people.images.shape[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1956, 3840, 3)\n",
      "(13233, 62, 47, 3)\n"
     ]
    }
   ],
   "source": [
    "print(file_.shape)\n",
    "print(patches.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='processing positive examples', max=13233), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\envs\\cw_env\\lib\\site-packages\\skimage\\feature\\_hog.py:119: skimage_deprecation: Default value of `block_norm`==`L1` is deprecated and will be changed to `L2-Hys` in v0.15\n",
      "  'be changed to `L2-Hys` in v0.15', skimage_deprecation)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "people_fv = list()\n",
    "for img in tqdm_notebook(people['images'], desc='processing positive examples'):\n",
    "    #visualise=True\n",
    "    fv, hog_img = hog(img[:, :, 2], orientations=8, pixels_per_cell=(6, 6), cells_per_block=(1, 1), feature_vector=True, visualise=True)\n",
    "    people_fv.append(fv)\n",
    "    # print(fv.shape)\n",
    "    # print(fv)\n",
    "    \"\"\"\n",
    "    fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(12, 6), sharex=True, sharey=True)\n",
    "    ax1.axis('off')\n",
    "    ax1.imshow(img.astype(np.int))\n",
    "    ax1.set_title('Input image')\n",
    "    ax2.axis('off')\n",
    "    ax2.imshow(hog_img, cmap=plt.cm.gray)\n",
    "    ax2.set_title('Histogram of Oriented Gradients')\n",
    "    plt.show()\n",
    "    \"\"\"\n",
    "people_fv=np.asarray(people_fv)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(560,)\n",
      "13233\n"
     ]
    }
   ],
   "source": [
    "print(people_fv[0].shape)\n",
    "print(len(people_fv))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='processing negative examples', max=13233), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\envs\\cw_env\\lib\\site-packages\\skimage\\feature\\_hog.py:119: skimage_deprecation: Default value of `block_norm`==`L1` is deprecated and will be changed to `L2-Hys` in v0.15\n",
      "  'be changed to `L2-Hys` in v0.15', skimage_deprecation)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "patch_fv = list()\n",
    "for patch in tqdm_notebook(patches, desc='processing negative examples'):\n",
    "    #print(patch.shape)\n",
    "    fv, hog_patch = hog(patch[:, :, 2], orientations=8, pixels_per_cell=(6, 6), cells_per_block=(1, 1), feature_vector=True, visualise=True)\n",
    "    patch_fv.append(fv)\n",
    "    \"\"\"\n",
    "    fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(12, 6), sharex=True, sharey=True)\n",
    "    ax1.axis('off')\n",
    "    ax1.imshow(patch)\n",
    "    ax1.set_title('Input image')\n",
    "    ax2.axis('off')\n",
    "    ax2.imshow(hog_patch,  cmap=plt.cm.gray)\n",
    "    ax2.set_title('Histogram of Oriented Gradients')\n",
    "    plt.show()\n",
    "    \"\"\"\n",
    "patch_fv = np.asarray(patch_fv)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(26466, 560)\n",
      "[[0.03987536 0.47275938 0.21401908 ... 0.01881485 0.03012654 0.05858849]\n",
      " [0.01425075 0.18887777 0.2577391  ... 0.02433732 0.01549511 0.19706225]\n",
      " [0.25256104 0.62938581 0.06237439 ... 0.         0.         0.        ]\n",
      " ...\n",
      " [0.18072718 0.24772575 0.31622968 ... 0.05177573 0.07189899 0.03585996]\n",
      " [0.16530757 0.04975588 0.15271736 ... 0.13763868 0.2053392  0.19413811]\n",
      " [0.02678954 0.05374218 0.06365722 ... 0.01416995 0.05112067 0.18943369]]\n",
      "(26466,)\n",
      "[1. 1. 1. ... 0. 0. 0.]\n",
      "0.5052138431313284\n",
      "0.49997281510078323\n"
     ]
    }
   ],
   "source": [
    "label=np.hstack((np.ones(len(people_fv)), np.zeros(len(patch_fv))))\n",
    "all_ex=np.vstack((people_fv, patch_fv))\n",
    "print(all_ex.shape)\n",
    "print(label.shape)\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(all_ex, label, shuffle=True, random_state=42)#42-ответ на все вопросы\n",
    "\n",
    "print(y_test.mean())\n",
    "print(y_test.std())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearSVC(C=1.0, class_weight=None, dual=True, fit_intercept=True,\n",
       "     intercept_scaling=1, loss='squared_hinge', max_iter=1000,\n",
       "     multi_class='ovr', penalty='l2', random_state=42, tol=0.0001,\n",
       "     verbose=0)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cls = LinearSVC(random_state=42)#Автостопом по галактике\n",
    "cls.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0., 0., 1., ..., 1., 0., 0.])"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_pred=cls.predict(X_test)\n",
    "x_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "score: 0.9919903279431767\n",
      "micro: 0.9919903279431767\n",
      "macro: 0.9919899064429678\n",
      "weighted: 0.9919905123495181\n",
      "None: [0.9919318  0.99204801]\n"
     ]
    }
   ],
   "source": [
    "print(f\"score: {cls.score(X_test, y_test)}\")\n",
    "avrg = ['micro', 'macro', 'weighted', None]\n",
    "for a in avrg:\n",
    "    print(f\"{a}: {f1_score(y_test, x_pred, average=a)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(62, 47)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "['cls.pkl']"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(people.images.shape[1:3])\n",
    "joblib.dump(cls, 'cls.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}