superResolution.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": [],
      "gpuType": "T4"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "source": [
        "#!pip install Tensorflow\n",
        "#!pip install Keras\n"
      ],
      "metadata": {
        "id": "yA42a56BJVxt"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "wZT6hmZdLBUM"
      },
      "outputs": [],
      "source": [
        "import scipy\n",
        "from glob import glob\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "from keras.datasets import cifar10\n",
        "from PIL import Image\n",
        "\n",
        "\n",
        "class DataLoader():\n",
        "    def __init__(self, dataset_name, img_res=(128,128)):\n",
        "        self.dataset_name = dataset_name\n",
        "        self.img_res = img_res\n",
        "\n",
        "    def load_data(self, batch_size=1, is_testing=False):\n",
        "        (x, y), (_, _) = cifar10.load_data()\n",
        "        x=np.asarray(x)\n",
        "        y=np.asarray(y)\n",
        "\n",
        "        data_type = \"train\" if not is_testing else \"test\"\n",
        "        batch_images = np.random.choice(range(x.shape[0]), size=batch_size)\n",
        "        imgs_hr = []\n",
        "        imgs_lr = []\n",
        "\n",
        "        for img_index in batch_images:\n",
        "            img = x[img_index, :, :, :]\n",
        "            h, w = self.img_res\n",
        "            low_h, low_w = int(h / 4), int(w / 4)\n",
        "\n",
        "            img_hr = Image.fromarray(img)\n",
        "            img_hr = img_hr.resize(self.img_res, Image.BICUBIC)\n",
        "            img_hr = np.array(img_hr)\n",
        "\n",
        "            img_lr = Image.fromarray(img)\n",
        "            img_lr = img_lr.resize((low_w, low_h), Image.BICUBIC)\n",
        "            img_lr = np.array(img_lr)\n",
        "\n",
        "            if not is_testing and np.random.random() < 0.5:\n",
        "                img_hr = np.fliplr(img_hr)\n",
        "                img_lr = np.fliplr(img_lr)\n",
        "\n",
        "            imgs_hr.append(img_hr)\n",
        "            imgs_lr.append(img_lr)\n",
        "\n",
        "        imgs_hr = np.array(imgs_hr) / 127.5 - 1\n",
        "        imgs_lr = np.array(imgs_lr) / 127.5 - 1\n",
        "\n",
        "        return imgs_hr, imgs_lr\n",
        "\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from __future__ import print_function, division\n",
        "import scipy\n",
        "from keras.layers import BatchNormalization, Input, Dense, Reshape, Flatten, Dropout, Concatenate, Activation, ZeroPadding2D, Add, Conv2D, UpSampling2D\n",
        "from tensorflow.keras.applications import VGG19\n",
        "from keras.models import Sequential, Model\n",
        "from tensorflow.keras.layers import LeakyReLU, PReLU\n",
        "from keras.optimizers.legacy import Adam\n",
        "import datetime\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "import sys\n",
        "import os\n",
        "\n",
        "\n",
        "import keras.backend as k\n"
      ],
      "metadata": {
        "id": "pa9zTwwyLEkN"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "class SRGAN():\n",
        "    def __init__(self):\n",
        "        self.channels=3\n",
        "        self.lr_height=64\n",
        "        self.lr_width=64\n",
        "\n",
        "\n",
        "        self.lr_shape=(self.lr_height, self.lr_width, self.channels)\n",
        "\n",
        "\n",
        "        self.hr_height=self.lr_height*4\n",
        "        self.hr_width=self.lr_width*4\n",
        "        self.hr_shape=(self.hr_height,self.hr_width,self.channels)\n",
        "\n",
        "        # Residual blocks\n",
        "\n",
        "        self.n_residual=16\n",
        "\n",
        "        optimizer=Adam(0.0002,0.5)\n",
        "\n",
        "        #pretrained VGG model to extract features from the high resolution and the generated high res images, min(mse)\n",
        "\n",
        "        self.vgg=self.build_vgg()\n",
        "        self.vgg.trainable=False\n",
        "        self.vgg.compile(loss='mse', optimizer=optimizer, metrics=['accuracy'])\n",
        "\n",
        "        #config Dataloader\n",
        "\n",
        "        self.dataset_name=\"cifar10_dataset\"\n",
        "        self.dataloader=DataLoader(dataset_name=self.dataset_name, img_res=(self.hr_height,self.hr_width))\n",
        "\n",
        "\n",
        "        patch=int(self.hr_height/2**4)\n",
        "        self.disc_patch=(patch,patch,1)\n",
        "\n",
        "        # Filters in gen and discriminator\n",
        "        self.gf=64\n",
        "        self.df=64\n",
        "\n",
        "        #build and compile descriminator\n",
        "\n",
        "        self.descriminator= self.build_descriminator()\n",
        "        self.descriminator.compile(loss='mse',optimizer=optimizer, metrics=['accuracy'])\n",
        "\n",
        "        #builld the generator\n",
        "\n",
        "        self.generator=self.build_generator()\n",
        "\n",
        "\n",
        "        #high res and low res images\n",
        "\n",
        "        img_hr=Input(shape=self.hr_shape)\n",
        "        img_lr=Input(shape=self.lr_shape)\n",
        "\n",
        "\n",
        "        #gen hr from lr\n",
        "\n",
        "        fake_hr=self.generator(img_lr)\n",
        "\n",
        "        #extract VGG19 features of this image\n",
        "\n",
        "        fake_features=self.vgg(fake_hr)\n",
        "\n",
        "        # in combined model gen is trainable and descriminator as non trainable\n",
        "\n",
        "        self.descriminator.trainable=False\n",
        "\n",
        "        validity=self.descriminator(fake_hr)\n",
        "        self.combined=Model([img_lr, img_hr],[validity, fake_features])\n",
        "        self.combined.compile(loss=['binary_crossentropy','mse'], loss_weights=[1e-3, 1], optimizer=optimizer)\n",
        "\n",
        "    def build_vgg(self):\n",
        "        vgg = VGG19(weights=\"imagenet\",include_top=False, input_shape=self.hr_shape)\n",
        "        return(Model(inputs=vgg.input,outputs=vgg.layers[9].output))\n",
        "\n",
        "\n",
        "    def build_generator(self):\n",
        "        def residual_block(layer_input, filters):\n",
        "            d=Conv2D(filters,kernel_size=3,strides=1, padding='same')(layer_input)\n",
        "            d=Activation('relu')(d)\n",
        "            d=BatchNormalization(momentum=.8)(d)\n",
        "            d=Conv2D(filters,kernel_size=3,strides=1, padding='same')(d)\n",
        "            d=Add()([d,layer_input])\n",
        "            return(d)\n",
        "\n",
        "        def deconv2D(layer_input):\n",
        "            u=UpSampling2D(size=2)(layer_input)\n",
        "            u=Conv2D(256,kernel_size=3, strides=1, padding='same')(u)\n",
        "            u=Activation('relu')(u)\n",
        "            return(u)\n",
        "\n",
        "        # lr image input\n",
        "        img_lr=Input(shape=self.lr_shape)\n",
        "\n",
        "        #pre residual block\n",
        "\n",
        "        c1=Conv2D(64, kernel_size=9, strides=1, padding='same')(img_lr)\n",
        "        c1=Activation('relu')(c1)\n",
        "\n",
        "\n",
        "        #Residual blocks\n",
        "\n",
        "        r=residual_block(c1,self.gf)\n",
        "        for _ in range(self.n_residual-1):\n",
        "            r=residual_block(r,self.gf)\n",
        "\n",
        "        #post residual blocks\n",
        "\n",
        "        c2=Conv2D(64, kernel_size=3, strides=1, padding='same')(r)\n",
        "        c2=BatchNormalization(momentum=.8)(c2)\n",
        "        c2=Add()([c2,c1])\n",
        "\n",
        "        #upsampling\n",
        "        u1=deconv2D(c2)\n",
        "        u2=deconv2D(u1)\n",
        "\n",
        "        #hr o/p\n",
        "\n",
        "        gen_hr=Conv2D(self.channels, kernel_size=9, strides=1, padding='same',activation='tanh')(u2)\n",
        "\n",
        "        return(Model(img_lr,gen_hr))\n",
        "\n",
        "    def build_descriminator(self):\n",
        "      def d_block(layer_input,filters,strides=1,bn=True):\n",
        "        #discriminator layer\n",
        "        d=Conv2D(filters,kernel_size=3,strides=strides,padding='same')(layer_input)\n",
        "        d=LeakyReLU(alpha=.2)(d)\n",
        "        if bn:\n",
        "          d=BatchNormalization(momentum=.8)(d)\n",
        "\n",
        "        return(d)\n",
        "\n",
        "      #input image\n",
        "      d0=Input(shape=self.hr_shape)\n",
        "\n",
        "      d1=d_block(d0,self.df,bn=False)\n",
        "      d2=d_block(d1,self.df,strides=2)\n",
        "      d3=d_block(d2,self.df*2)\n",
        "      d4=d_block(d3,self.df*2,strides=2)\n",
        "      d5=d_block(d4,self.df*4)\n",
        "\n",
        "      d6=d_block(d5,self.df*4,strides=2)\n",
        "      d7=d_block(d6,self.df*8)\n",
        "      d8=d_block(d7,self.df*8,strides=2)\n",
        "\n",
        "      d9=Dense(self.df*16)(d8)\n",
        "      d10=LeakyReLU(alpha=.2)(d9)\n",
        "      validity=Dense(1,activation='sigmoid')(d10)\n",
        "\n",
        "      return(Model(d0,validity))\n",
        "\n",
        "    def train(self,epochs,batch_size=1,sample_intervals=50):\n",
        "\n",
        "      start_time=datetime.datetime.now()\n",
        "\n",
        "      for epoch in range(epochs):\n",
        "        #train discriminator\n",
        "\n",
        "\n",
        "        #sample images and their conditioning counterparts\n",
        "        imgs_hr,imgs_lr=self.dataloader.load_data(batch_size)\n",
        "\n",
        "        # lr to hr (fake)\n",
        "\n",
        "        fake_hr=self.generator.predict(imgs_lr)\n",
        "\n",
        "        valid=np.ones((batch_size,)+self.disc_patch)\n",
        "        fake=np.zeros((batch_size,)+self.disc_patch)\n",
        "\n",
        "\n",
        "        #train the discriminator\n",
        "\n",
        "        d_loss_real=self.descriminator.train_on_batch(imgs_hr,valid)\n",
        "        d_loss_fake=self.descriminator.train_on_batch(fake_hr,fake)\n",
        "\n",
        "        d_loss=.5*np.add(d_loss_real,d_loss_fake)\n",
        "\n",
        "        #Train gen\n",
        "\n",
        "        imgs_hr, imgs_lr=self.dataloader.load_data(batch_size)\n",
        "\n",
        "        valid=np.ones((batch_size,)+self.disc_patch)\n",
        "\n",
        "        image_features=self.vgg.predict(imgs_hr)\n",
        "\n",
        "        g_loss = self.combined.train_on_batch([imgs_lr, imgs_hr], [valid, image_features])\n",
        "        elapsed_time=datetime.datetime.now()-start_time\n",
        "\n",
        "        #plotting\n",
        "        print(\"%d time: %s\"%(epoch,elapsed_time))\n",
        "\n",
        "        if epoch%sample_intervals==0:\n",
        "          self.sample_images(epoch)\n",
        "\n",
        "\n",
        "    def sample_images(self,epoch):\n",
        "      os.makedirs(\"images/%s\"%self.dataset_name,exist_ok=True)\n",
        "\n",
        "      r,c=2,2\n",
        "\n",
        "      imgs_hr,imgs_lr=self.dataloader.load_data(batch_size=2,is_testing=True)\n",
        "\n",
        "      fake_hr=self.generator.predict(imgs_lr)\n",
        "\n",
        "      imgs_lr=.5*imgs_lr+.5\n",
        "      fake_hr=.5*fake_hr+.5\n",
        "      imgs_hr=.5*imgs_hr+.5\n",
        "\n",
        "\n",
        "      titles=['Generated','Original']\n",
        "\n",
        "      fig,axs=plt.subplots(r,c)\n",
        "\n",
        "      cnt=0\n",
        "\n",
        "      for row in range(r):\n",
        "        for col, image in enumerate([fake_hr,imgs_hr]):\n",
        "          axs[row,col].imshow(image[row])\n",
        "          axs[row,col].set_title(titles[col])\n",
        "          axs[row,col].axis(\"off\")\n",
        "        cnt+=1\n",
        "      plt.savefig('images/%s/%d.png'%(self.dataset_name,epoch))\n",
        "      plt.close()\n",
        "\n",
        "      for i in range(r):\n",
        "        fig=plt.figure()\n",
        "        plt.imshow(imgs_lr[i])\n",
        "        fig.savefig('images/%s/%d_lowres%d.png'%(self.dataset_name,epoch,i))\n",
        "        plt.close()\n",
        "\n",
        "    def save_model(self):\n",
        "      def save(model,model_name):\n",
        "        model_path=\"saved_model/%s.json\"%model_name\n",
        "        weights_path=\"saved_model/%s_weights.hdf5\"%model_name\n",
        "        options={'file_arch':model_path,\n",
        "                 'file_weight':weights_path}\n",
        "        json_string=model.to_json()\n",
        "        open(options['file_arch'],'w').write(json_string)\n",
        "        model.save_weights(options['file_weight'])\n",
        "\n",
        "      save(self.generator,\"generator\")\n",
        "      save(self.descriminator,\"discriminator\")\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n"
      ],
      "metadata": {
        "id": "2LjhNg7gLLQU"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "if __name__==\"__main__\":\n",
        "  gan=SRGAN()\n",
        "  gan.train(epochs =100, batch_size=1, sample_intervals=10)\n",
        "  !mkdir saved_model\n",
        "  gan.save_model()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XABN1FOX_fwa",
        "outputId": "6811fffa-907d-4037-f30e-23eb0eab480c"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/vgg19/vgg19_weights_tf_dim_ordering_tf_kernels_notop.h5\n",
            "80134624/80134624 [==============================] - 1s 0us/step\n",
            "Downloading data from https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz\n",
            "170498071/170498071 [==============================] - 2s 0us/step\n",
            "1/1 [==============================] - 10s 10s/step\n",
            "1/1 [==============================] - 0s 423ms/step\n",
            "0 time: 0:00:30.098856\n",
            "1/1 [==============================] - 2s 2s/step\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "1 time: 0:00:36.099395\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "2 time: 0:00:38.406297\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "3 time: 0:00:40.457463\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "4 time: 0:00:42.344154\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "1/1 [==============================] - 0s 27ms/step\n",
            "5 time: 0:00:44.965628\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "6 time: 0:00:46.965499\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "7 time: 0:00:48.929651\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "8 time: 0:00:50.782159\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 24ms/step\n",
            "9 time: 0:00:52.737100\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "10 time: 0:00:54.619690\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "11 time: 0:00:58.926984\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "12 time: 0:01:00.793631\n",
            "1/1 [==============================] - 0s 24ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "13 time: 0:01:02.670620\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 24ms/step\n",
            "14 time: 0:01:04.594272\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "15 time: 0:01:06.534547\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "1/1 [==============================] - 0s 30ms/step\n",
            "16 time: 0:01:09.113640\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "17 time: 0:01:11.293706\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "18 time: 0:01:13.330051\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "19 time: 0:01:15.286194\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "20 time: 0:01:17.197599\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "21 time: 0:01:20.959230\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "22 time: 0:01:23.316634\n",
            "1/1 [==============================] - 0s 42ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "23 time: 0:01:26.427262\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "24 time: 0:01:28.374732\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "25 time: 0:01:30.315283\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "26 time: 0:01:32.409836\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "27 time: 0:01:34.979072\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "28 time: 0:01:36.949657\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "29 time: 0:01:38.910986\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "30 time: 0:01:40.788567\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "31 time: 0:01:44.330184\n",
            "1/1 [==============================] - 0s 32ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "32 time: 0:01:47.083830\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "33 time: 0:01:48.947288\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "34 time: 0:01:50.810530\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "35 time: 0:01:52.734587\n",
            "1/1 [==============================] - 0s 25ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "36 time: 0:01:54.607337\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "37 time: 0:01:56.724495\n",
            "1/1 [==============================] - 0s 45ms/step\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "38 time: 0:01:59.454937\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "39 time: 0:02:01.436905\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "40 time: 0:02:03.375468\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "41 time: 0:02:06.714171\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "42 time: 0:02:08.665992\n",
            "1/1 [==============================] - 0s 36ms/step\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "43 time: 0:02:11.543156\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "44 time: 0:02:13.486896\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "45 time: 0:02:15.436051\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "46 time: 0:02:17.287699\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "47 time: 0:02:19.123762\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "48 time: 0:02:21.082365\n",
            "1/1 [==============================] - 0s 34ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "49 time: 0:02:23.862735\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "50 time: 0:02:25.733672\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "1/1 [==============================] - 0s 25ms/step\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "51 time: 0:02:29.214337\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "52 time: 0:02:31.149837\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "53 time: 0:02:33.127512\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "54 time: 0:02:35.913786\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "55 time: 0:02:37.773114\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "56 time: 0:02:39.873100\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "57 time: 0:02:41.836503\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "58 time: 0:02:43.842384\n",
            "1/1 [==============================] - 0s 24ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "59 time: 0:02:45.894627\n",
            "1/1 [==============================] - 0s 39ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "60 time: 0:02:48.599897\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 25ms/step\n",
            "61 time: 0:02:52.237786\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "62 time: 0:02:54.221002\n",
            "1/1 [==============================] - 0s 25ms/step\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "63 time: 0:02:56.189625\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "1/1 [==============================] - 0s 28ms/step\n",
            "64 time: 0:02:58.308888\n",
            "1/1 [==============================] - 0s 37ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "65 time: 0:03:00.942993\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "66 time: 0:03:02.918954\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "67 time: 0:03:04.851841\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "68 time: 0:03:06.717296\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "69 time: 0:03:08.677123\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "1/1 [==============================] - 0s 29ms/step\n",
            "70 time: 0:03:10.907968\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "71 time: 0:03:14.818011\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "72 time: 0:03:16.690162\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "73 time: 0:03:18.536243\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "74 time: 0:03:20.488453\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "75 time: 0:03:22.387947\n",
            "1/1 [==============================] - 0s 33ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "76 time: 0:03:25.150415\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 24ms/step\n",
            "77 time: 0:03:27.105974\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "78 time: 0:03:28.991049\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "79 time: 0:03:30.991336\n",
            "1/1 [==============================] - 0s 27ms/step\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "80 time: 0:03:32.938283\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 36ms/step\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "81 time: 0:03:37.158229\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "82 time: 0:03:39.045581\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "83 time: 0:03:41.007638\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "84 time: 0:03:42.947581\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "85 time: 0:03:44.892774\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "1/1 [==============================] - 0s 40ms/step\n",
            "86 time: 0:03:46.833633\n",
            "1/1 [==============================] - 0s 31ms/step\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "87 time: 0:03:49.591469\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "88 time: 0:03:51.423075\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "89 time: 0:03:53.308371\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "90 time: 0:03:55.252448\n",
            "1/1 [==============================] - 0s 25ms/step\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "91 time: 0:03:58.643215\n",
            "1/1 [==============================] - 0s 37ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "92 time: 0:04:01.441121\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "93 time: 0:04:03.377123\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 23ms/step\n",
            "94 time: 0:04:05.258693\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 18ms/step\n",
            "95 time: 0:04:07.128442\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "96 time: 0:04:09.108693\n",
            "1/1 [==============================] - 0s 26ms/step\n",
            "1/1 [==============================] - 0s 21ms/step\n",
            "97 time: 0:04:11.112080\n",
            "1/1 [==============================] - 0s 35ms/step\n",
            "1/1 [==============================] - 0s 20ms/step\n",
            "98 time: 0:04:13.947113\n",
            "1/1 [==============================] - 0s 22ms/step\n",
            "1/1 [==============================] - 0s 19ms/step\n",
            "99 time: 0:04:15.900719\n"
          ]
        }
      ]
    }
  ]
}