deep_dream.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from __future__ import print_function\n",
    "\n",
    "import os\n",
    "import time\n",
    "\n",
    "import numpy as np\n",
    "\n",
    "from matplotlib import pylab as plt\n",
    "import cv2\n",
    "\n",
    "from scipy.misc import imsave\n",
    "from scipy.optimize import fmin_l_bfgs_b\n",
    "\n",
    "from keras.preprocessing.image import load_img, img_to_array\n",
    "from keras.applications import vgg16\n",
    "from keras import backend as K\n",
    "from keras.layers import Input"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "base_image_path = os.path.join(\"data\",\"devfest_gdg_300x300.jpg\")\n",
    "result_prefix = os.path.join(\"data\",\"dream_devfest\")\n",
    "\n",
    "# dimensions of the generated picture.\n",
    "img_width = 300\n",
    "img_height = 300\n",
    "\n",
    "# some settings we found interesting\n",
    "saved_settings = {\n",
    "    'cool_trip': {'features': {'block1_conv1': 0.05,\n",
    "                              'block2_conv2': 0.05,\n",
    "                              'block3_conv3': 0.05},\n",
    "                 'continuity': 0.1,\n",
    "                 'dream_l2': 0.3,\n",
    "                 'jitter': 5},\n",
    "    'bad_trip': {'features': {'block4_conv1': 0.05,\n",
    "                              'block4_conv2': 0.01,\n",
    "                              'block4_conv3': 0.01},\n",
    "                 'continuity': 0.1,\n",
    "                 'dream_l2': 0.8,\n",
    "                 'jitter': 5},\n",
    "    'dreamy': {'features': {'block5_conv1': 0.05,\n",
    "                            'block5_conv2': 0.02},\n",
    "               'continuity': 0.1,\n",
    "               'dream_l2': 0.02,\n",
    "               'jitter': 0},\n",
    "}\n",
    "# the settings we will use in this experiment\n",
    "settings = saved_settings['cool_trip']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def preprocess_image(image_path):\n",
    "    img = load_img(image_path, target_size=(img_width, img_height))\n",
    "    img = img_to_array(img)\n",
    "    img = np.expand_dims(img, axis=0)\n",
    "    img = vgg16.preprocess_input(img)\n",
    "    return img\n",
    "\n",
    "# util function to convert a tensor into a valid image\n",
    "def deprocess_image(x):\n",
    "    x = x.reshape((img_width, img_height, 3))\n",
    "    # Remove zero-center by mean pixel\n",
    "    x[:, :, 0] += 103.939\n",
    "    x[:, :, 1] += 116.779\n",
    "    x[:, :, 2] += 123.68\n",
    "    # 'BGR'->'RGB'\n",
    "    x = x[:, :, ::-1]\n",
    "    x = np.clip(x, 0, 255).astype('uint8')\n",
    "    return x\n",
    "\n",
    "img_size = (img_width, img_height, 3)\n",
    "# this will contain our generated image\n",
    "dream = Input(batch_shape=(1,) + img_size)\n",
    "\n",
    "# build the VGG16 network with our placeholder\n",
    "# the model will be loaded with pre-trained ImageNet weights\n",
    "model = vgg16.VGG16(input_tensor=dream,\n",
    "                    weights='imagenet', include_top=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# get the symbolic outputs of each \"key\" layer (we gave them unique names).\n",
    "layer_dict = dict([(layer.name, layer) for layer in model.layers])\n",
    "\n",
    "# continuity loss util function\n",
    "def continuity_loss(x):\n",
    "    assert K.ndim(x) == 4\n",
    "    a = K.square(x[:, :img_width - 1, :img_height-1, :] -\n",
    "                 x[:, 1:, :img_height - 1, :])\n",
    "    b = K.square(x[:, :img_width - 1, :img_height-1, :] -\n",
    "                 x[:, :img_width - 1, 1:, :])\n",
    "    return K.sum(K.pow(a + b, 1.25))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# define the loss\n",
    "loss = K.variable(0.)\n",
    "for layer_name in settings['features']:\n",
    "    # add the L2 norm of the features of a layer to the loss\n",
    "    assert layer_name in layer_dict.keys(), 'Layer ' + layer_name + ' not found in model.'\n",
    "    coeff = settings['features'][layer_name]\n",
    "    x = layer_dict[layer_name].output\n",
    "    shape = layer_dict[layer_name].output_shape\n",
    "    # we avoid border artifacts by only involving non-border pixels in the loss\n",
    "    loss -= coeff * K.sum(K.square(x[:, 2: shape[1] - 2, 2: shape[2] - 2, :])) / np.prod(shape[1:])\n",
    "\n",
    "# add continuity loss (gives image local coherence, can result in an artful blur)\n",
    "loss += settings['continuity'] * continuity_loss(dream) / np.prod(img_size)\n",
    "# add image L2 norm to loss (prevents pixels from taking very high values, makes image darker)\n",
    "loss += settings['dream_l2'] * K.sum(K.square(dream)) / np.prod(img_size)\n",
    "\n",
    "# compute the gradients of the dream wrt the loss\n",
    "grads = K.gradients(loss, dream)\n",
    "\n",
    "outputs = [loss]\n",
    "if type(grads) in {list, tuple}:\n",
    "    outputs += grads\n",
    "else:\n",
    "    outputs.append(grads)\n",
    "\n",
    "f_outputs = K.function([dream], outputs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1, 300, 300, 64)\n",
      "(1, 75, 75, 256)\n",
      "(1, 150, 150, 128)\n"
     ]
    }
   ],
   "source": [
    "def eval_loss_and_grads(x):\n",
    "    x = x.reshape((1,) + img_size)\n",
    "    outs = f_outputs([x])\n",
    "    loss_value = outs[0]\n",
    "    if len(outs[1:]) == 1:\n",
    "        grad_values = outs[1].flatten().astype('float64')\n",
    "    else:\n",
    "        grad_values = np.array(outs[1:]).flatten().astype('float64')\n",
    "    return loss_value, grad_values\n",
    "\n",
    "# this Evaluator class makes it possible\n",
    "# to compute loss and gradients in one pass\n",
    "# while retrieving them via two separate functions,\n",
    "# \"loss\" and \"grads\". This is done because scipy.optimize\n",
    "# requires separate functions for loss and gradients,\n",
    "# but computing them separately would be inefficient.\n",
    "class Evaluator(object):\n",
    "    def __init__(self):\n",
    "        self.loss_value = None\n",
    "        self.grad_values = None\n",
    "\n",
    "    def loss(self, x):\n",
    "        assert self.loss_value is None\n",
    "        loss_value, grad_values = eval_loss_and_grads(x)\n",
    "        self.loss_value = loss_value\n",
    "        self.grad_values = grad_values\n",
    "        return self.loss_value\n",
    "\n",
    "    def grads(self, x):\n",
    "        assert self.loss_value is not None\n",
    "        grad_values = np.copy(self.grad_values)\n",
    "        self.loss_value = None\n",
    "        self.grad_values = None\n",
    "        return grad_values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": []
    }
   ],
   "source": [
    "evaluator = Evaluator()\n",
    "\n",
    "# run scipy-based optimization (L-BFGS) over the pixels of the generated image\n",
    "# so as to minimize the loss\n",
    "x = preprocess_image(base_image_path)\n",
    "\n",
    "for i in range(15):\n",
    "    print('Start of iteration', i)\n",
    "    start_time = time.time()\n",
    "\n",
    "    # add a random jitter to the initial image. This will be reverted at decoding time\n",
    "    random_jitter = (settings['jitter'] * 2) * (np.random.random(img_size) - 0.5)\n",
    "    x += random_jitter\n",
    "\n",
    "    # run L-BFGS for 7 steps\n",
    "    x, min_val, info = fmin_l_bfgs_b(evaluator.loss, x.flatten(),\n",
    "                                     fprime=evaluator.grads, maxfun=7)\n",
    "    print('Current loss value:', min_val)\n",
    "    # decode the dream and save it\n",
    "    x = x.reshape(img_size)\n",
    "    x -= random_jitter\n",
    "    img = deprocess_image(np.copy(x))\n",
    "    plt.imsave(os.path.join(\"results\",\"dog_dream_%s.jpg\"%i),img)\n",
    "    plt.imshow(img)\n",
    "    plt.show()\n",
    "    end_time = time.time()\n",
    "    print('Iteration %d completed in %ds' % (i, end_time - start_time))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "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": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}