keras_train_model_inception-grayscale.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Another version of the CNN to load files in greyscale before tuning the inception model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "import src.picture_stuff as pix\n",
    "\n",
    "import os\n",
    "#import tensorflow as tf\n",
    "import cPickle as pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.applications.inception_v3 import InceptionV3\n",
    "from keras.preprocessing import image\n",
    "from keras.models import Model\n",
    "from keras.layers import Dense, GlobalAveragePooling2D\n",
    "from keras import backend as K\n",
    "\n",
    "# create the base pre-trained model\n",
    "base_model = InceptionV3(weights='imagenet', include_top=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "image_default_size = 299\n",
    "use_gray = True"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "###  Start here with import of pickled photo files -\n",
    "see  \n",
    "http://localhost:8888/notebooks/CAPSTONE/lego-database/app_stuff/Process_pictures_pickle.ipynb"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "pkl_path = '../brick_pics_pickled/'\n",
    "\n",
    "with open(pkl_path + \"X.pkl\") as f_un:\n",
    "    X = pickle.load(f_un)\n",
    "\n",
    "with open(pkl_path + \"y_list.pkl\") as f_un:\n",
    "    y_list = pickle.load(f_un)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Convert saved files to grayscale\n",
    "if use_gray:\n",
    "    X = pix.convert_to_gray(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1665, 299, 299, 3)\n"
     ]
    }
   ],
   "source": [
    "print X.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "label_dic,y_out = pix.y_to_hot(y_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{0: 2357, 1: 2412, 2: 2413, 3: 2420, 4: 2431, 5: 2450, 6: 2780, 7: 3001, 8: 3002, 9: 3003, 10: 3004, 11: 3005, 12: 3007, 13: 3009, 14: 3010, 15: 3020, 16: 3022, 17: 3023, 18: 3024, 19: 3030, 20: 3031, 21: 3032, 22: 3037, 23: 3039, 24: 3062, 25: 3069, 26: 3070, 27: 3297, 28: 3298, 29: 3308, 30: 3460, 31: 3622, 32: 3660, 33: 3666, 34: 3710, 35: 3795, 36: 3895, 37: 4445, 38: 4743, 39: 4854, 40: 4861, 41: 6141, 42: 6179, 43: 6183, 44: 6215, 45: 30499, 46: 30503, 47: 30504, 48: 41748, 49: 41769, 50: 44237, 51: 48183, 52: 48933, 53: 50304, 54: 50955, 55: 61487}\n"
     ]
    }
   ],
   "source": [
    "print label_dic"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# current working directory\n",
    "filepath = os.getcwd()\n",
    "\n",
    "if use_gray:\n",
    "    fname = filepath + \"/saved_models/label_dic_all-gray.pkl\"\n",
    "else:\n",
    "    fname = filepath + \"/saved_models/label_dic_all.pkl\"\n",
    "\n",
    "with open(fname, 'w') as f:\n",
    "    pickle.dump(label_dic, f)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### NOTE: IN FINAL VERSION, LET THE WHOLE X AND Y SETS BECOME THE TRAINING SETS, AND WORK ONLY ON NEW PICTURES FOR CLASSIFIER"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# TRAINING ON ALL OR DOING TRAIN-TEST SPLIT\n",
    "\n",
    "#X_train, X_test, y_train, y_test = train_test_split( X, y_out, test_size=0.25, random_state=42)\n",
    "X_train = X\n",
    "y_train = y_out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1665, 299, 299, 3)\n",
      "(1665, 56)\n"
     ]
    }
   ],
   "source": [
    "print X_train.shape\n",
    "print y_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "#stop here"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### OK to here -- we have training data.  Now to the keras stuff\n",
    "Almost directly from  \n",
    "https://keras.io/applications/#inceptionv3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# add a global spatial average pooling layer\n",
    "x = base_model.output\n",
    "x = GlobalAveragePooling2D()(x)\n",
    "# let's add a fully-connected layer\n",
    "x = Dense(1024, activation='relu')(x)\n",
    "# and a logistic layer -- one node per class (width of y_hot)\n",
    "num_classes = y_train.shape[1]\n",
    "predictions = Dense(num_classes, activation='softmax')(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# this is the model we will train\n",
    "model = Model(inputs=base_model.input, outputs=predictions)\n",
    "\n",
    "# first: train only the top layers (which were randomly initialized)\n",
    "# i.e. freeze all convolutional InceptionV3 layers\n",
    "for layer in base_model.layers:\n",
    "    layer.trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# compile the model (should be done *after* setting layers to non-trainable)\n",
    "model.compile(optimizer='rmsprop', loss='categorical_crossentropy',metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      " 608/1665 [=========>....................] - ETA: 316s - loss: 5.0221 - acc: 0.0263\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-16-92b579ec8a94>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m# NOTE: ran this cell several times\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX_train\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0my_train\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mepochs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m10\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch_size\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m32\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m/Users/deborahwilliams/anaconda/lib/python2.7/site-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, **kwargs)\u001b[0m\n\u001b[1;32m   1596\u001b[0m                               \u001b[0minitial_epoch\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0minitial_epoch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1597\u001b[0m                               \u001b[0msteps_per_epoch\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0msteps_per_epoch\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1598\u001b[0;31m                               validation_steps=validation_steps)\n\u001b[0m\u001b[1;32m   1599\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1600\u001b[0m     def evaluate(self, x, y,\n",
      "\u001b[0;32m/Users/deborahwilliams/anaconda/lib/python2.7/site-packages/keras/engine/training.pyc\u001b[0m in \u001b[0;36m_fit_loop\u001b[0;34m(self, f, ins, out_labels, batch_size, epochs, verbose, callbacks, val_f, val_ins, shuffle, callback_metrics, initial_epoch, steps_per_epoch, validation_steps)\u001b[0m\n\u001b[1;32m   1181\u001b[0m                     \u001b[0mbatch_logs\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'size'\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbatch_ids\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1182\u001b[0m                     \u001b[0mcallbacks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mon_batch_begin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mbatch_index\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch_logs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1183\u001b[0;31m                     \u001b[0mouts\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mf\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mins_batch\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1184\u001b[0m                     \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0misinstance\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mouts\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlist\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1185\u001b[0m                         \u001b[0mouts\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mouts\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/Users/deborahwilliams/anaconda/lib/python2.7/site-packages/keras/backend/tensorflow_backend.pyc\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, inputs)\u001b[0m\n\u001b[1;32m   2271\u001b[0m         updated = session.run(self.outputs + [self.updates_op],\n\u001b[1;32m   2272\u001b[0m                               \u001b[0mfeed_dict\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2273\u001b[0;31m                               **self.session_kwargs)\n\u001b[0m\u001b[1;32m   2274\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mupdated\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0moutputs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   2275\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/Users/deborahwilliams/anaconda/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36mrun\u001b[0;34m(self, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m    776\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    777\u001b[0m       result = self._run(None, fetches, feed_dict, options_ptr,\n\u001b[0;32m--> 778\u001b[0;31m                          run_metadata_ptr)\n\u001b[0m\u001b[1;32m    779\u001b[0m       \u001b[0;32mif\u001b[0m \u001b[0mrun_metadata\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    780\u001b[0m         \u001b[0mproto_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtf_session\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mTF_GetBuffer\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mrun_metadata_ptr\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/Users/deborahwilliams/anaconda/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_run\u001b[0;34m(self, handle, fetches, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m    980\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mfinal_fetches\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mfinal_targets\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    981\u001b[0m       results = self._do_run(handle, final_targets, final_fetches,\n\u001b[0;32m--> 982\u001b[0;31m                              feed_dict_string, options, run_metadata)\n\u001b[0m\u001b[1;32m    983\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    984\u001b[0m       \u001b[0mresults\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/Users/deborahwilliams/anaconda/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_do_run\u001b[0;34m(self, handle, target_list, fetch_list, feed_dict, options, run_metadata)\u001b[0m\n\u001b[1;32m   1030\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mhandle\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1031\u001b[0m       return self._do_call(_run_fn, self._session, feed_dict, fetch_list,\n\u001b[0;32m-> 1032\u001b[0;31m                            target_list, options, run_metadata)\n\u001b[0m\u001b[1;32m   1033\u001b[0m     \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1034\u001b[0m       return self._do_call(_prun_fn, self._session, handle, feed_dict,\n",
      "\u001b[0;32m/Users/deborahwilliams/anaconda/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_do_call\u001b[0;34m(self, fn, *args)\u001b[0m\n\u001b[1;32m   1037\u001b[0m   \u001b[0;32mdef\u001b[0m \u001b[0m_do_call\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1038\u001b[0m     \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1039\u001b[0;31m       \u001b[0;32mreturn\u001b[0m \u001b[0mfn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0margs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1040\u001b[0m     \u001b[0;32mexcept\u001b[0m \u001b[0merrors\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mOpError\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1041\u001b[0m       \u001b[0mmessage\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mcompat\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mas_text\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0me\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmessage\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/Users/deborahwilliams/anaconda/lib/python2.7/site-packages/tensorflow/python/client/session.pyc\u001b[0m in \u001b[0;36m_run_fn\u001b[0;34m(session, feed_dict, fetch_list, target_list, options, run_metadata)\u001b[0m\n\u001b[1;32m   1019\u001b[0m         return tf_session.TF_Run(session, options,\n\u001b[1;32m   1020\u001b[0m                                  \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtarget_list\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1021\u001b[0;31m                                  status, run_metadata)\n\u001b[0m\u001b[1;32m   1022\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1023\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0m_prun_fn\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0msession\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhandle\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfeed_dict\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfetch_list\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "# NOTE: ran this cell several times\n",
    "model.fit(X_train,y_train, epochs=10, batch_size=32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def save_my_model(save_dir,model_name):\n",
    "    '''Save current Keras model and weights\n",
    "       INPUTS: save_dir:   STRING, desired directory\n",
    "               model_name: STRING, desired file name\n",
    "       OUTPUTS: none\n",
    "    '''\n",
    "    if not os.path.isdir(save_dir):\n",
    "        os.makedirs(save_dir)\n",
    "    model_path = os.path.join(save_dir, model_name)\n",
    "    model.save(model_path)\n",
    "    print('Saved model at %s ' % model_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# SAVE THE PARTIALLY TRAINED MODEL\n",
    "\n",
    "save_dir = os.path.join(os.getcwd(), 'saved_models')\n",
    "\n",
    "if use_gray:\n",
    "    model_name = 'keras_inception_all_partially_trained_gray.h5'\n",
    "else:\n",
    "    model_name = 'keras_inception_all_partially_trained.h5'\n",
    "\n",
    "save_my_model(save_dir,model_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Number of model layers:\n",
    "print len(base_model.layers)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# we chose to train the top 2 inception blocks, i.e. we will freeze\n",
    "# the first 249 layers and unfreeze the rest:\n",
    "for layer in model.layers[:249]:\n",
    "   layer.trainable = False\n",
    "for layer in model.layers[249:]:\n",
    "   layer.trainable = True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# we need to recompile the model for these modifications to take effect\n",
    "# we use SGD with a low learning rate\n",
    "from keras.optimizers import SGD\n",
    "model.compile(optimizer=SGD(lr=0.0001, momentum=0.9),\n",
    "              loss='categorical_crossentropy',metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# we train our model again (this time fine-tuning the top 2 inception blocks\n",
    "# alongside the top Dense layers\n",
    "#model.fit_generator(...)\n",
    "\n",
    "model.fit(X_train,y_train, epochs=10, batch_size=32)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Once we have a working model, save, evaluate, etc.  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# SAVE THE MODEL\n",
    "save_dir = os.path.join(os.getcwd(), 'saved_models')\n",
    "\n",
    "if use_gray:\n",
    "    model_name = 'keras_inception_all_trained-gray.h5'\n",
    "else:\n",
    "    model_name = 'keras_inception_all_trained.h5'\n",
    "\n",
    "save_my_model(save_dir,model_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "#stop here"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "train_more = True\n",
    "if train_more:\n",
    "    model.fit(X_train,y_train, epochs=30, batch_size=32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "if train_more:\n",
    "    # SAVE THE MODEL AGAIN\n",
    "    save_dir = os.path.join(os.getcwd(), 'saved_models')\n",
    "    if use_gray:\n",
    "        model_name = 'keras_inception_all_re-trained-grey.h5'\n",
    "    else:\n",
    "        model_name = 'keras_inception_all_re-trained.h5'\n",
    "    save_my_model(save_dir,model_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "stop here"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Cheating: testing on training data\n",
    "model.test_on_batch(X_train,y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Cheating: testing on training data\n",
    "predict_gen = model.predict_on_batch(X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "num_predictions = 10\n",
    "\n",
    "for predict_index, predicted_y in enumerate(predict_gen):\n",
    "    # STILL CHEATING\n",
    "    #actual_label = label_dic[np.argmax(y_test[predict_index])]\n",
    "    actual_label = label_dic[np.argmax(y_train[predict_index])]\n",
    "    predicted_label = label_dic[np.argmax(predicted_y)]\n",
    "    print('Actual Label = %s vs. Predicted Label = %s' % (actual_label,\n",
    "                                                          predicted_label))\n",
    "    if predict_index == num_predictions:\n",
    "        break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "model.test_on_batch(X_test,y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "predict_gen = model.predict_on_batch(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "num_predictions = 10\n",
    "for predict_index, predicted_y in enumerate(predict_gen):\n",
    "    actual_label = label_dic[np.argmax(y_test[predict_index])]\n",
    "    predicted_label = label_dic[np.argmax(predicted_y)]\n",
    "    print('Actual Label = %s vs. Predicted Label = %s' % (actual_label,\n",
    "                                                          predicted_label))\n",
    "    if predict_index == num_predictions:\n",
    "        break"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### PART 4D: make predictions on a single picture / label combination"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "reload_model = False\n",
    "if reload_model:\n",
    "    from keras.models import load_model\n",
    "    # reload the most recently saved model and train more\n",
    "    filepath = '/Users/deborahwilliams/Documents/Galvanize/DSI_class/CAPSTONE/lego-database/'\n",
    "    fname = filepath + 'saved_models/keras_example_inception_trained-gray.h5'\n",
    "    model = load_model(fname)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "if reload_model:\n",
    "    filepath = os.getcwd()\n",
    "    with open(filepath + \"/saved_models/label_dic-gray.pkl\") as f_un:\n",
    "        label_dic = pickle.load(f_un)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "picture_index_lookup = {}\n",
    "for idx,label in enumerate(y_list):\n",
    "    picture_index_lookup[label]=idx"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "idx0 = 101\n",
    "small_X_test = X_test[idx0:idx0+1]\n",
    "small_y_test = [y_test[idx0]]\n",
    "\n",
    "small_predict_gen = model.predict_on_batch(small_X_test)\n",
    "\n",
    "small_prediction_list,small_prediction_weights = pix.make_full_prediction_list(\n",
    "    small_predict_gen,label_dic,n_match=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "actual = small_y_test[0]\n",
    "preds = small_prediction_list[0]\n",
    "\n",
    "idx_preds = [picture_index_lookup[pred] for pred in preds]\n",
    "#act_pic = X[picture_index_lookup[label_dic[np.argmax(actual)]]]\n",
    "act_pic = small_X_test\n",
    "print \"label for this item {}\".format(actual)\n",
    "print \"Top 5 predictions \", preds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "fig, ax = plt.subplots(1,figsize=(8,8))\n",
    "\n",
    "ax1 = plt.subplot2grid((4, 4), (0, 0), colspan=3, rowspan=4)\n",
    "ax1.imshow(act_pic[0])\n",
    "ax1.set_title(\"new brick pic\")\n",
    "ax1.grid(False)\n",
    "ax1.axis('off')\n",
    "\n",
    "for idx1 in range(4):\n",
    "    ax2 = plt.subplot2grid((4, 4), (idx1, 3))\n",
    "    ax2.imshow(X[idx_preds[idx1]])\n",
    "    ax2.set_title(\"predict #{}, wt {:.3f}\".format(idx1,small_prediction_weights[0][idx1]))\n",
    "    ax2.grid(False)\n",
    "    ax2.axis('off')\n",
    "#fig.savefig(\"temp.png\")"
   ]
  }
 ],
 "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": "2.7.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}