NeuralStyleTransfer.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "currdir = os.getcwd()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'D:\\\\Projects\\\\MLProjects'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "currdir"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.preprocessing.image import load_img,img_to_array\n",
    "\n",
    "target_image_path = os.path.join(currdir,'potrait.jpg')\n",
    "style_reference_image_path = os.path.join(currdir,'transfer_style_reference.jpg')\n",
    "\n",
    "width , height = load_img(target_image_path).size\n",
    "img_height = 400\n",
    "img_width = int(width*img_height/height)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from keras.applications import vgg19\n",
    "\n",
    "def preprocess_image(image_path):\n",
    "    img = load_img(image_path,target_size=(img_height,img_width))\n",
    "    img = img_to_array(img)\n",
    "    img = np.expand_dims(img,axis = 0)\n",
    "    img = vgg19.preprocess_input(img)\n",
    "    return img\n",
    "\n",
    "def deprocess_image(x):\n",
    "    x[:,:,0] += 103.939    ##  these 3 steps are used for the mean\n",
    "    x[:,:,1] += 116.779    ##  centering of the pixel values\n",
    "    x[:,:,2] += 123.68     ##  of the given image\n",
    "    x = x[:,:,::-1]      # this step tranforms the image from BGR format to RGB format\n",
    "    x = np.clip(x,0,255).astype('uint8')\n",
    "    return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras import backend as K\n",
    "\n",
    "target_image = K.constant(preprocess_image(target_image_path))\n",
    "style_reference_image = K.constant(preprocess_image(style_reference_image_path))\n",
    "combination_image = K.placeholder((1,img_height,img_width,3))\n",
    "\n",
    "input_tensor = K.concatenate([target_image,style_reference_image,combination_image] , axis = 0)\n",
    "\n",
    "model = vgg19.VGG19(input_tensor = input_tensor,weights = 'imagenet',include_top = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def content_loss(base,combination):\n",
    "    return K.sum(K.square(combination - base))\n",
    "\n",
    "def gram_matrix(x):\n",
    "    features = K.batch_flatten(K.permute_dimensions(x,(2,0,1)))\n",
    "    gram = K.dot(features,K.transpose(features))\n",
    "    return gram\n",
    "\n",
    "def style_loss(style,combination):\n",
    "    S = gram_matrix(style)\n",
    "    C = gram_matrix(combination)\n",
    "    channels = 3\n",
    "    size = img_height*img_width\n",
    "    return K.sum(K.square(S-C))/(4.*(channels**2)*(size**2))\n",
    "\n",
    "def total_variation_loss(x):\n",
    "    a = K.square( x[:,:img_height-1,:img_width-1,:] - x[:,1:,:img_width-1,:] )\n",
    "    b = K.square( x[:,:img_height-1,:img_width-1,:] - x[:,1:,:img_height-1,:] )\n",
    "    return K.sum(K.pow(a+b,1.25))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "outputs_dict = {layer.name:layer.output for layer in model.layers}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'input_2': <tf.Tensor 'concat_1:0' shape=(3, 400, 400, 3) dtype=float32>,\n",
       " 'block1_conv1': <tf.Tensor 'block1_conv1_1/Relu:0' shape=(3, 400, 400, 64) dtype=float32>,\n",
       " 'block1_conv2': <tf.Tensor 'block1_conv2_1/Relu:0' shape=(3, 400, 400, 64) dtype=float32>,\n",
       " 'block1_pool': <tf.Tensor 'block1_pool_1/MaxPool:0' shape=(3, 200, 200, 64) dtype=float32>,\n",
       " 'block2_conv1': <tf.Tensor 'block2_conv1_1/Relu:0' shape=(3, 200, 200, 128) dtype=float32>,\n",
       " 'block2_conv2': <tf.Tensor 'block2_conv2_1/Relu:0' shape=(3, 200, 200, 128) dtype=float32>,\n",
       " 'block2_pool': <tf.Tensor 'block2_pool_1/MaxPool:0' shape=(3, 100, 100, 128) dtype=float32>,\n",
       " 'block3_conv1': <tf.Tensor 'block3_conv1_1/Relu:0' shape=(3, 100, 100, 256) dtype=float32>,\n",
       " 'block3_conv2': <tf.Tensor 'block3_conv2_1/Relu:0' shape=(3, 100, 100, 256) dtype=float32>,\n",
       " 'block3_conv3': <tf.Tensor 'block3_conv3_1/Relu:0' shape=(3, 100, 100, 256) dtype=float32>,\n",
       " 'block3_conv4': <tf.Tensor 'block3_conv4_1/Relu:0' shape=(3, 100, 100, 256) dtype=float32>,\n",
       " 'block3_pool': <tf.Tensor 'block3_pool_1/MaxPool:0' shape=(3, 50, 50, 256) dtype=float32>,\n",
       " 'block4_conv1': <tf.Tensor 'block4_conv1_1/Relu:0' shape=(3, 50, 50, 512) dtype=float32>,\n",
       " 'block4_conv2': <tf.Tensor 'block4_conv2_1/Relu:0' shape=(3, 50, 50, 512) dtype=float32>,\n",
       " 'block4_conv3': <tf.Tensor 'block4_conv3_1/Relu:0' shape=(3, 50, 50, 512) dtype=float32>,\n",
       " 'block4_conv4': <tf.Tensor 'block4_conv4_1/Relu:0' shape=(3, 50, 50, 512) dtype=float32>,\n",
       " 'block4_pool': <tf.Tensor 'block4_pool_1/MaxPool:0' shape=(3, 25, 25, 512) dtype=float32>,\n",
       " 'block5_conv1': <tf.Tensor 'block5_conv1_1/Relu:0' shape=(3, 25, 25, 512) dtype=float32>,\n",
       " 'block5_conv2': <tf.Tensor 'block5_conv2_1/Relu:0' shape=(3, 25, 25, 512) dtype=float32>,\n",
       " 'block5_conv3': <tf.Tensor 'block5_conv3_1/Relu:0' shape=(3, 25, 25, 512) dtype=float32>,\n",
       " 'block5_conv4': <tf.Tensor 'block5_conv4_1/Relu:0' shape=(3, 25, 25, 512) dtype=float32>,\n",
       " 'block5_pool': <tf.Tensor 'block5_pool_1/MaxPool:0' shape=(3, 12, 12, 512) dtype=float32>}"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "outputs_dict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "content_layer = 'block5_conv2'\n",
    "style_layers = [\n",
    "                'block1_conv1',\n",
    "                'block2_conv1',\n",
    "                'block3_conv1',\n",
    "                'block4_conv1',\n",
    "                'block5_conv1'\n",
    "                ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "total_variation_weight = 1e-4\n",
    "style_weight = 1.0\n",
    "content_weight = 0.025"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "loss = K.variable(0.0)\n",
    "layer_features = outputs_dict[content_layer]\n",
    "target_image_features = layer_features[0,:,:,:]\n",
    "combination_features = layer_features[2,:,:,:]\n",
    "loss = loss + content_weight*content_loss(target_image_features,combination_features)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "for layer_name in style_layers:\n",
    "    layer_features = outputs_dict[layer_name]\n",
    "    style_reference_features = layer_features[1,:,:,:]\n",
    "    combination_features = layer_features[2,:,:,:]\n",
    "    s1 = style_loss(style_reference_features,combination_features)\n",
    "    loss += (style_weight/len(style_layers))*s1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "loss += total_variation_weight*total_variation_loss(combination_image)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "grads = K.gradients(loss,combination_image)[0]\n",
    "fetch_loss_and_grads = K.function([combination_image],[loss,grads])\n",
    "\n",
    "class Evaluator(object):\n",
    "    \n",
    "    def __init__(self):\n",
    "        self.loss_value = None\n",
    "        self.grads_value = None\n",
    "    def loss(self,x):\n",
    "        assert self.loss_value is None\n",
    "        x = x.reshape((1,img_height,img_width,3))\n",
    "        outs = fetch_loss_and_grads([x])\n",
    "        loss_value = outs[0]\n",
    "        grads_value = outs[1].flatten().astype('float64')\n",
    "        self.loss_value = loss_value\n",
    "        self.grads_value = grads_value\n",
    "        return self.loss_value\n",
    "    def grads(self,x):\n",
    "        assert self.loss_value is not None\n",
    "        grads_value = np.copy(self.grads_value)\n",
    "        self.loss_value = None\n",
    "        self.grads_value = None\n",
    "        return grads_value\n",
    "evaluator = Evaluator()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Start of iteration  0\n",
      "Current loss value: 13670634000.0\n",
      "Image saved as neural_style_result_at_iteration_0.png\n",
      "Iteration 0 completed in 11s\n",
      "Start of iteration  1\n",
      "Current loss value: 7092416000.0\n",
      "Image saved as neural_style_result_at_iteration_1.png\n",
      "Iteration 1 completed in 11s\n",
      "Start of iteration  2\n",
      "Current loss value: 5106887700.0\n",
      "Image saved as neural_style_result_at_iteration_2.png\n",
      "Iteration 2 completed in 11s\n",
      "Start of iteration  3\n",
      "Current loss value: 4176121300.0\n",
      "Image saved as neural_style_result_at_iteration_3.png\n",
      "Iteration 3 completed in 11s\n",
      "Start of iteration  4\n",
      "Current loss value: 3561225200.0\n",
      "Image saved as neural_style_result_at_iteration_4.png\n",
      "Iteration 4 completed in 11s\n",
      "Start of iteration  5\n",
      "Current loss value: 3156382000.0\n",
      "Image saved as neural_style_result_at_iteration_5.png\n",
      "Iteration 5 completed in 11s\n",
      "Start of iteration  6\n",
      "Current loss value: 2754625300.0\n",
      "Image saved as neural_style_result_at_iteration_6.png\n",
      "Iteration 6 completed in 11s\n",
      "Start of iteration  7\n",
      "Current loss value: 2453814300.0\n",
      "Image saved as neural_style_result_at_iteration_7.png\n",
      "Iteration 7 completed in 11s\n",
      "Start of iteration  8\n",
      "Current loss value: 2166610700.0\n",
      "Image saved as neural_style_result_at_iteration_8.png\n",
      "Iteration 8 completed in 11s\n",
      "Start of iteration  9\n",
      "Current loss value: 1946727200.0\n",
      "Image saved as neural_style_result_at_iteration_9.png\n",
      "Iteration 9 completed in 11s\n",
      "Start of iteration  10\n",
      "Current loss value: 1745333200.0\n",
      "Image saved as neural_style_result_at_iteration_10.png\n",
      "Iteration 10 completed in 12s\n",
      "Start of iteration  11\n",
      "Current loss value: 1594421800.0\n",
      "Image saved as neural_style_result_at_iteration_11.png\n",
      "Iteration 11 completed in 11s\n",
      "Start of iteration  12\n",
      "Current loss value: 1466649300.0\n",
      "Image saved as neural_style_result_at_iteration_12.png\n",
      "Iteration 12 completed in 12s\n",
      "Start of iteration  13\n",
      "Current loss value: 1359636600.0\n",
      "Image saved as neural_style_result_at_iteration_13.png\n",
      "Iteration 13 completed in 13s\n",
      "Start of iteration  14\n",
      "Current loss value: 1270501500.0\n",
      "Image saved as neural_style_result_at_iteration_14.png\n",
      "Iteration 14 completed in 11s\n",
      "Start of iteration  15\n",
      "Current loss value: 1188515300.0\n",
      "Image saved as neural_style_result_at_iteration_15.png\n",
      "Iteration 15 completed in 12s\n",
      "Start of iteration  16\n",
      "Current loss value: 1108970200.0\n",
      "Image saved as neural_style_result_at_iteration_16.png\n",
      "Iteration 16 completed in 12s\n",
      "Start of iteration  17\n",
      "Current loss value: 1045418300.0\n",
      "Image saved as neural_style_result_at_iteration_17.png\n",
      "Iteration 17 completed in 12s\n",
      "Start of iteration  18\n",
      "Current loss value: 984517760.0\n",
      "Image saved as neural_style_result_at_iteration_18.png\n",
      "Iteration 18 completed in 12s\n",
      "Start of iteration  19\n",
      "Current loss value: 930961860.0\n",
      "Image saved as neural_style_result_at_iteration_19.png\n",
      "Iteration 19 completed in 12s\n",
      "Start of iteration  20\n",
      "Current loss value: 876257400.0\n",
      "Image saved as neural_style_result_at_iteration_20.png\n",
      "Iteration 20 completed in 12s\n",
      "Start of iteration  21\n",
      "Current loss value: 829547900.0\n",
      "Image saved as neural_style_result_at_iteration_21.png\n",
      "Iteration 21 completed in 12s\n",
      "Start of iteration  22\n",
      "Current loss value: 785785660.0\n",
      "Image saved as neural_style_result_at_iteration_22.png\n",
      "Iteration 22 completed in 12s\n",
      "Start of iteration  23\n",
      "Current loss value: 737240260.0\n",
      "Image saved as neural_style_result_at_iteration_23.png\n",
      "Iteration 23 completed in 12s\n",
      "Start of iteration  24\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-28-b5a4187950df>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[0;32m     10\u001b[0m     \u001b[0mprint\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m'Start of iteration '\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mi\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     11\u001b[0m     \u001b[0mstart_time\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtime\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 12\u001b[1;33m     \u001b[0mx\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mmin_val\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0minfo\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mfmin_l_bfgs_b\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mevaluator\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mloss\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mfprime\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mevaluator\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mgrads\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mmaxfun\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;36m20\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     13\u001b[0m     \u001b[0mprint\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m'Current loss value:'\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mmin_val\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     14\u001b[0m     \u001b[0mimg\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mx\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mcopy\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mreshape\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mimg_height\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mimg_width\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m3\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\envs\\tf-gpu\\lib\\site-packages\\scipy\\optimize\\lbfgsb.py\u001b[0m in \u001b[0;36mfmin_l_bfgs_b\u001b[1;34m(func, x0, fprime, args, approx_grad, bounds, m, factr, pgtol, epsilon, iprint, maxfun, maxiter, disp, callback, maxls)\u001b[0m\n\u001b[0;32m    196\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    197\u001b[0m     res = _minimize_lbfgsb(fun, x0, args=args, jac=jac, bounds=bounds,\n\u001b[1;32m--> 198\u001b[1;33m                            **opts)\n\u001b[0m\u001b[0;32m    199\u001b[0m     d = {'grad': res['jac'],\n\u001b[0;32m    200\u001b[0m          \u001b[1;34m'task'\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0mres\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;34m'message'\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\envs\\tf-gpu\\lib\\site-packages\\scipy\\optimize\\lbfgsb.py\u001b[0m in \u001b[0;36m_minimize_lbfgsb\u001b[1;34m(fun, x0, args, jac, bounds, disp, maxcor, ftol, gtol, eps, maxfun, maxiter, iprint, callback, maxls, finite_diff_rel_step, **unknown_options)\u001b[0m\n\u001b[0;32m    358\u001b[0m             \u001b[1;31m# until the completion of the current minimization iteration.\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    359\u001b[0m             \u001b[1;31m# Overwrite f and g:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 360\u001b[1;33m             \u001b[0mf\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mg\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mfunc_and_grad\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    361\u001b[0m         \u001b[1;32melif\u001b[0m \u001b[0mtask_str\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mstartswith\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34mb'NEW_X'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    362\u001b[0m             \u001b[1;31m# new iteration\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\envs\\tf-gpu\\lib\\site-packages\\scipy\\optimize\\_differentiable_functions.py\u001b[0m in \u001b[0;36mfun_and_grad\u001b[1;34m(self, x)\u001b[0m\n\u001b[0;32m    198\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[1;32mnot\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0marray_equal\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    199\u001b[0m             \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_update_x_impl\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 200\u001b[1;33m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_update_fun\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    201\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_update_grad\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    202\u001b[0m         \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mf\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mg\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\envs\\tf-gpu\\lib\\site-packages\\scipy\\optimize\\_differentiable_functions.py\u001b[0m in \u001b[0;36m_update_fun\u001b[1;34m(self)\u001b[0m\n\u001b[0;32m    164\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0m_update_fun\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    165\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[1;32mnot\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mf_updated\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 166\u001b[1;33m             \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_update_fun_impl\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    167\u001b[0m             \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mf_updated\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mTrue\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    168\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\envs\\tf-gpu\\lib\\site-packages\\scipy\\optimize\\_differentiable_functions.py\u001b[0m in \u001b[0;36mupdate_fun\u001b[1;34m()\u001b[0m\n\u001b[0;32m     71\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     72\u001b[0m         \u001b[1;32mdef\u001b[0m \u001b[0mupdate_fun\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 73\u001b[1;33m             \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mf\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mfun_wrapped\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     74\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     75\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_update_fun_impl\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mupdate_fun\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\envs\\tf-gpu\\lib\\site-packages\\scipy\\optimize\\_differentiable_functions.py\u001b[0m in \u001b[0;36mfun_wrapped\u001b[1;34m(x)\u001b[0m\n\u001b[0;32m     68\u001b[0m         \u001b[1;32mdef\u001b[0m \u001b[0mfun_wrapped\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     69\u001b[0m             \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mnfev\u001b[0m \u001b[1;33m+=\u001b[0m \u001b[1;36m1\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 70\u001b[1;33m             \u001b[1;32mreturn\u001b[0m \u001b[0mfun\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     71\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     72\u001b[0m         \u001b[1;32mdef\u001b[0m \u001b[0mupdate_fun\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m<ipython-input-26-fad2514dd15d>\u001b[0m in \u001b[0;36mloss\u001b[1;34m(self, x)\u001b[0m\n\u001b[0;32m     10\u001b[0m         \u001b[1;32massert\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mloss_value\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     11\u001b[0m         \u001b[0mx\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mx\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mreshape\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mimg_height\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mimg_width\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;36m3\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 12\u001b[1;33m         \u001b[0mouts\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mfetch_loss_and_grads\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     13\u001b[0m         \u001b[0mloss_value\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mouts\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     14\u001b[0m         \u001b[0mgrads_value\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mouts\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mflatten\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mastype\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m'float64'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\envs\\tf-gpu\\lib\\site-packages\\tensorflow_core\\python\\keras\\backend.py\u001b[0m in \u001b[0;36m__call__\u001b[1;34m(self, inputs)\u001b[0m\n\u001b[0;32m   3733\u001b[0m     return nest.pack_sequence_as(\n\u001b[0;32m   3734\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_outputs_structure\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 3735\u001b[1;33m         \u001b[1;33m[\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_numpy\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[1;32min\u001b[0m \u001b[0moutputs\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m  \u001b[1;31m# pylint: disable=protected-access\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   3736\u001b[0m         expand_composites=True)\n\u001b[0;32m   3737\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\envs\\tf-gpu\\lib\\site-packages\\tensorflow_core\\python\\keras\\backend.py\u001b[0m in \u001b[0;36m<listcomp>\u001b[1;34m(.0)\u001b[0m\n\u001b[0;32m   3733\u001b[0m     return nest.pack_sequence_as(\n\u001b[0;32m   3734\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_outputs_structure\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 3735\u001b[1;33m         \u001b[1;33m[\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_numpy\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[1;32min\u001b[0m \u001b[0moutputs\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m  \u001b[1;31m# pylint: disable=protected-access\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   3736\u001b[0m         expand_composites=True)\n\u001b[0;32m   3737\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\envs\\tf-gpu\\lib\\site-packages\\tensorflow_core\\python\\framework\\ops.py\u001b[0m in \u001b[0;36m_numpy\u001b[1;34m(self)\u001b[0m\n\u001b[0;32m    906\u001b[0m     \u001b[1;31m# pylint: disable=protected-access\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    907\u001b[0m     \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 908\u001b[1;33m       \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_numpy_internal\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    909\u001b[0m     \u001b[1;32mexcept\u001b[0m \u001b[0mcore\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_NotOkStatusException\u001b[0m \u001b[1;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    910\u001b[0m       \u001b[0msix\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mraise_from\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mcore\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_status_to_exception\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0me\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mcode\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0me\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mmessage\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "from scipy.optimize import fmin_l_bfgs_b\n",
    "from keras.preprocessing.image import save_img\n",
    "import time\n",
    "\n",
    "result_prefix = 'neural_style_result'\n",
    "iterations = 100\n",
    "x = preprocess_image(target_image_path)\n",
    "x = x.flatten()\n",
    "for i in range(iterations):\n",
    "    print('Start of iteration ',i)\n",
    "    start_time = time.time()\n",
    "    x,min_val,info = fmin_l_bfgs_b(evaluator.loss,x,fprime=evaluator.grads,maxfun = 20)\n",
    "    print('Current loss value:', min_val)\n",
    "    img = x.copy().reshape((img_height, img_width, 3))\n",
    "    img = deprocess_image(img)\n",
    "    fname = result_prefix + '_at_iteration_%d.png' % i\n",
    "    save_img(fname, img)\n",
    "    print('Image saved as', fname)\n",
    "    end_time = time.time()\n",
    "    print('Iteration %d completed in %ds' % (i, end_time - start_time))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "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": "3.6.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}