Visual Embeddings.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import nltk\n",
    "from nltk.corpus import wordnet as wn\n",
    "from gensim.models import KeyedVectors\n",
    "from gensim.models import KeyedVectors\n",
    "import scipy\n",
    "import cv2\n",
    "from keras.applications.resnet50 import ResNet50\n",
    "from google_images_download import google_images_download\n",
    "from sklearn.manifold import TSNE"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Get list of nouns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Numer of nouns: 8\n"
     ]
    }
   ],
   "source": [
    "nouns = set()\n",
    "for synset in list(wn.all_synsets('n'))[:10]:\n",
    "\n",
    "    n = synset.name().split('.')[0]\n",
    "\n",
    "    if len(n) > 2 and n.isalpha():\n",
    "        nouns.add(n)\n",
    "        \n",
    "print 'Numer of nouns:', len(nouns)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Create Word Vectors"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Completed Image ====> 1. oops-concept-abstraction-793x397.jpg\n",
      "\n",
      "Errors: 0\n",
      "\n",
      "Completed Image ====> 1. screen_shot_2016-01-18_at_12.25.34_pm_109860.jpg\n",
      "\n",
      "Errors: 0\n",
      "\n",
      "Completed Image ====> 1. 1*0ttfuc9a8rmcqxcbrnemng.png\n",
      "\n",
      "Errors: 0\n",
      "\n",
      "Completed Image ====> 1. mv5byjizmzc4mgitytnkzi00zmyyltlhztgtyzqxyzewngy2nmuyxkeyxkfqcgdeqxvynzq5mzy0njm@._v1_uy268_cr2,0,182,268_al_.jpg\n",
      "\n",
      "Errors: 0\n",
      "\n",
      "Completed Image ====> 1. latest?cb=20150223191205.jpg\n",
      "\n",
      "Errors: 0\n",
      "\n",
      "Completed Image ====> 1. congener_levels.jpg\n",
      "\n",
      "Errors: 0\n",
      "\n",
      "Completed Image ====> 1. whole-clipart-1.jpg\n",
      "\n",
      "Errors: 0\n",
      "\n",
      "Completed Image ====> 1. starfishnoaa.jpg\n",
      "\n",
      "Errors: 0\n",
      "\n",
      "Vocabulary size: 8\n"
     ]
    }
   ],
   "source": [
    "resnet = ResNet50(weights='imagenet', \n",
    "                  include_top=False, \n",
    "                  pooling='avg')\n",
    "embeddings = KeyedVectors(2048)\n",
    "for word in nouns:\n",
    "    response = google_images_download.googleimagesdownload()\n",
    "    path = response.download({'keywords': word, 'limit': 1})[word][0]\n",
    "    img = cv2.imread(path)\n",
    "    img = scipy.misc.imresize(img, 224.0 / img.shape[0])\n",
    "    img = img.reshape((1,) + img.shape)\n",
    "    embeddings[word] = resnet.predict(img)[0]\n",
    "    \n",
    "print 'Vocabulary size:', len(embeddings.vocab)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# labels: 8 # vectors: 8\n"
     ]
    }
   ],
   "source": [
    "labels, vectors = zip(*map(lambda v: (v, embeddings[v]), embeddings.vocab))\n",
    "print '# labels:', len(labels), '# vectors:', len(vectors)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[t-SNE] Computing 7 nearest neighbors...\n",
      "[t-SNE] Indexed 8 samples in 0.000s...\n",
      "[t-SNE] Computed neighbors for 8 samples in 0.001s...\n",
      "[t-SNE] Computed conditional probabilities for sample 8 / 8\n",
      "[t-SNE] Mean sigma: 0.209599\n",
      "[t-SNE] KL divergence after 250 iterations with early exaggeration: 60.322514\n",
      "[t-SNE] Error after 850 iterations: 0.130330\n"
     ]
    }
   ],
   "source": [
    "tsne = TSNE(n_components=2, n_iter=1000, perplexity=3, metric='cosine', verbose=1)\n",
    "transformed = tsne.fit_transform(vectors)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmIAAAJCCAYAAABj+qvQAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIv8fB6QDsxgzFxcAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x139dfb890>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "x = transformed[:,0]\n",
    "y = transformed[:,1]\n",
    "\n",
    "plt.figure(figsize=(10, 10)) \n",
    "for i in range(len(x)):\n",
    "    plt.scatter(x[i],y[i])\n",
    "    plt.annotate(labels[i],\n",
    "                 xy=(x[i], y[i]),\n",
    "                 xytext=(5, 2),\n",
    "                 textcoords='offset points',\n",
    "                 ha='right',\n",
    "                 va='bottom')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Translate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def translate(word):\n",
    "    response = google_images_download.googleimagesdownload()\n",
    "    path = response.download({'keywords': word, 'limit': 1})[word][0]\n",
    "    img = cv2.imread(path)\n",
    "    img = scipy.misc.imresize(img, 224.0 / img.shape[0])\n",
    "    vector = resnet.predict(img.reshape((1,) + img.shape))[0]\n",
    "    return embeddings.most_similar([vector])[0][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Completed Image ====> 1. ristorante-a-mano.jpg\n",
      "\n",
      "Errors: 0\n",
      "\n"
     ]
    }
   ],
   "source": [
    "translated = translate('ristorante')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "entity\n"
     ]
    }
   ],
   "source": [
    "print translated"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python (hiredscore)",
   "language": "python",
   "name": "hiredscore"
  },
  "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
}