3-Make Pipeline.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "source": [
    "Here I finalize the pipeline and ensure it gives me the\n",
    "probabilities I want at the end when given an image."
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "outputs": [],
   "source": [
    "import pickle\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import skimage\n",
    "import skimage.color\n",
    "import skimage.io\n",
    "import skimage.feature\n",
    "import skimage.transform\n",
    "%matplotlib inline"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "We transform the image into a mode where our model can process it.\n",
    "Then we rescale it by multiplying by 255.\n",
    "Then convert it to uint8 since our model expects an 8 bit array image.\n",
    "\n",
    "Then we convert it to grayscale\n",
    "\n",
    "Then into a hog feature."
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "outputs": [],
   "source": [
    "#pipeline model\n",
    "image = skimage.io.imread('eagle.jpg')\n",
    "\n",
    "#Transform image into 80 x 80\n",
    "image_resize = skimage.transform.resize(image, (80, 80))\n",
    "image_scale = 255*image_resize\n",
    "image_transform = image_scale.astype(np.uint8)\n",
    "\n",
    "#rgb to gray\n",
    "gray = skimage.color.rgb2gray(image_transform)\n",
    "#hog feature\n",
    "feature_vector = skimage.feature.hog(gray,\n",
    "                                     orientations=9,\n",
    "                                     pixels_per_cell=(8,8), cells_per_block=(3,3))\n",
    "\n",
    "\n"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "outputs": [
    {
     "data": {
      "text/plain": "array([0.01566118, 0.01359198, 0.02500137, ..., 0.03805931, 0.02104542,\n       0.00313196])"
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feature_vector"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "We then load our model and try to predict what the image is."
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "outputs": [],
   "source": [
    "model_svm = pickle.load(open('dsa_image_classification_svc.pickle', 'rb'))"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "outputs": [],
   "source": [
    "scaler = pickle.load(open('dsa_scaler.pickle', 'rb'))"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "outputs": [
    {
     "data": {
      "text/plain": "{'C': 1.0,\n 'break_ties': False,\n 'cache_size': 200,\n 'class_weight': None,\n 'coef0': 0.0,\n 'decision_function_shape': 'ovr',\n 'degree': 3,\n 'gamma': 'scale',\n 'kernel': 'rbf',\n 'max_iter': -1,\n 'probability': True,\n 'random_state': None,\n 'shrinking': True,\n 'tol': 0.001,\n 'verbose': False}"
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_svm.get_params()"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "For just a test, the model predicted eagle correctly."
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "outputs": [],
   "source": [
    "scalex = scaler.transform(feature_vector.reshape(1,-1))\n",
    "result = model_svm.predict(scalex)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "outputs": [
    {
     "data": {
      "text/plain": "array(['eagle'], dtype='<U8')"
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "outputs": [],
   "source": [
    "#probability\n",
    "predicted_prob = model_svm.predict_proba(scalex)\n",
    "predicted_prob = predicted_prob.flatten()\n",
    "labels = model_svm.classes_"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "outputs": [
    {
     "data": {
      "text/plain": "<BarContainer object of 20 artists>"
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": "<Figure size 432x288 with 1 Axes>",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.barh(labels, predicted_prob)\n"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "outputs": [
    {
     "data": {
      "text/plain": "array([3.71710380e-04, 1.32500199e-03, 3.78408445e-03, 3.11653383e-03,\n       2.80152632e-04, 1.97183162e-03, 2.52398104e-02, 9.24167497e-01,\n       2.43734461e-04, 5.84007259e-04, 1.04753824e-04, 1.97603422e-03,\n       6.10250844e-04, 5.18944401e-04, 5.76324527e-03, 2.66491830e-02,\n       1.34743221e-03, 1.18137603e-03, 6.86017486e-04, 7.83983018e-05])"
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predicted_prob"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "outputs": [
    {
     "data": {
      "text/plain": "array(['bear', 'cat', 'chicken', 'cow', 'deer', 'dog', 'duck', 'eagle',\n       'elephant', 'human', 'lion', 'monkey', 'mouse', 'natural', 'panda',\n       'pigeon', 'rabbit', 'sheep', 'tiger', 'wolf'], dtype='<U8')"
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labels"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "outputs": [],
   "source": [
    "top_5_prob_ind = predicted_prob.argsort()[::-1][:5]"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "outputs": [
    {
     "data": {
      "text/plain": "array([ 7, 15,  6, 14,  2], dtype=int64)"
     },
     "execution_count": 79,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "top_5_prob_ind"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "outputs": [],
   "source": [
    "top_labels = labels[top_5_prob_ind]\n",
    "top_prob = predicted_prob[top_5_prob_ind]"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "markdown",
   "source": [
    "The probabilities of prediction"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "outputs": [],
   "source": [
    "top_dict = dict()\n",
    "for key, val in zip(top_labels, top_prob):\n",
    "    top_dict.update({key: str(np.round(val,2)*100) + \" %\"})\n",
    "#print(str(key) +\" \" + str(np.round(val,2)*100) + \" %\")"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "outputs": [],
   "source": [
    "def pipeline_model(path, scaler_transform, model):\n",
    "    #pipeline model\n",
    "    image = skimage.io.imread(path)\n",
    "\n",
    "    #Transform image into 80 x 80\n",
    "    image_resize = skimage.transform.resize(image, (80, 80))\n",
    "    image_scale = 255*image_resize\n",
    "    image_transform = image_scale.astype(np.uint8)\n",
    "\n",
    "    #rgb to gray\n",
    "    gray = skimage.color.rgb2gray(image_transform)\n",
    "    #hog feature\n",
    "    feature_vector = skimage.feature.hog(gray,\n",
    "                                         orientations=9,\n",
    "                                         pixels_per_cell=(8,8), cells_per_block=(3,3))\n",
    "\n",
    "    #scaling\n",
    "    scalex = scaler_transform.transform(feature_vector.reshape(1,-1))\n",
    "    result = model.predict(scalex)\n",
    "\n",
    "    #probability\n",
    "    predicted_prob = model.predict_proba(scalex)\n",
    "    predicted_prob = predicted_prob.flatten()\n",
    "    labels = model.classes_\n",
    "\n",
    "    top_5_prob_ind = predicted_prob.argsort()[::-1][:5]\n",
    "    top_labels = labels[top_5_prob_ind]\n",
    "    top_prob = predicted_prob[top_5_prob_ind]\n",
    "\n",
    "    top_dict = dict()\n",
    "    for key, val in zip(top_labels, top_prob):\n",
    "        top_dict.update({key: str(np.round(val,2)*100) + \" %\"})\n",
    "\n",
    "    return top_dict"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "outputs": [
    {
     "data": {
      "text/plain": "{'eagle': '92.0 %',\n 'pigeon': '3.0 %',\n 'duck': '3.0 %',\n 'panda': '1.0 %',\n 'chicken': '0.0 %'}"
     },
     "execution_count": 111,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pipeline_model('eagle.jpg', scaler, model_svm)"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%%\n"
    }
   }
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "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": 0
}