facial_cnn.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "facial_cnn.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "machine_shape": "hm"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "I3Y-FAsXffGD",
        "colab_type": "code",
        "outputId": "be7c0d5a-091b-4714-d0b8-128946cd30a1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "# Verify GPU\n",
        "import tensorflow as tf \n",
        "tf.test.gpu_device_name()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'/device:GPU:0'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JAW4po9Vi7yc",
        "colab_type": "code",
        "outputId": "cb1036de-f5e2-46b2-80bc-2bda73f9e511",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 33
        }
      },
      "source": [
        "# Connect to images folder and set constants\n",
        "from google.colab import drive\n",
        "drive.mount('/content/drive')\n",
        "\n"
      ],
      "execution_count": 34,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_cECUw6Sgioa",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Data processing\n",
        "import pandas as pd\n",
        "import cv2\n",
        "\n",
        "\n",
        "def dataset():\n",
        "    images = []\n",
        "    labels = []\n",
        "    df = pd.read_csv(LABELS)\n",
        "    for idx, row in df.iterrows():\n",
        "      print(\"[process] \", idx)\n",
        "      filepath = IMAGES_DIR + row['filename']\n",
        "      label = row['score']\n",
        "      image = cv2.imread(filepath)\n",
        "      try:\n",
        "        image = cv2.resize(image, (227, 227))\n",
        "      except Exception as e:\n",
        "        # In case image's size is smaller than 227 * 227\n",
        "        pass\n",
        "      images.append(image)\n",
        "      labels.append(label)\n",
        "    return images, labels\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pLwc00s4gMeD",
        "colab_type": "text"
      },
      "source": [
        "### *Training here*"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ATqQ5-u6JMWu",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import keras\n",
        "from keras.datasets import mnist\n",
        "from keras.models import Sequential\n",
        "from keras.models import model_from_json\n",
        "from keras.optimizers import Adam\n",
        "from keras.layers import (Dense, Dropout, Flatten, Activation,\n",
        "                          Conv2D, MaxPooling2D, BatchNormalization)\n",
        "from keras.layers.core import Activation\n",
        "from keras import backend as K\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "import numpy as np\n",
        "from sklearn.model_selection import train_test_split\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "\n",
        "import pandas as pd\n",
        "import pickle\n",
        "import os\n",
        "import datetime\n",
        "\n",
        "DIR = \"/content/drive/My Drive/Facial_beauty/\"\n",
        "IMAGES_DIR = DIR + \"Images/\"\n",
        "LABELS = 'https://raw.githubusercontent.com/ptran1203/facial_beauty/master/dataset/all_labels.csv'\n",
        "BATCH_SIZE = 64\n",
        "EPOCH = 100\n",
        "TEST_SIZE = 0.15\n",
        "VAL_SIZE = 0.15\n",
        "FIG_SIZE = (10,7)\n",
        "\n",
        "\n",
        "def pickle_save(object, path):\n",
        "    with open(path, \"wb\") as f:\n",
        "        return pickle.dump(object, f)\n",
        "\n",
        "\n",
        "def pickle_load(path):\n",
        "    with open(path, \"rb\") as f:\n",
        "        return pickle.load(f)\n",
        "\n",
        "def path(filename):\n",
        "    return os.path.join(DIR, filename)\n",
        "\n",
        "# Define path\n",
        "model_path = path('model.json')\n",
        "model_weight_path = path('model_weight.h5')\n",
        "history_path = path('history.pkl')\n",
        "\n",
        "# Load numpy images from drive\n",
        "labels_np = pickle_load(os.path.join(DIR, \"labels_np.pkl\"))\n",
        "imgs_np = pickle_load(os.path.join(DIR, \"images_np.pkl\"))\n",
        "\n",
        "# split train/val\n",
        "(trainX, valX, trainY, valY) = train_test_split(imgs_np,\n",
        "\tlabels_np, test_size=VAL_SIZE, random_state=42)\n",
        "# split train/test\n",
        "(trainX, testX, trainY, testY) = train_test_split(trainX,\n",
        "\ttrainY, test_size=TEST_SIZE, random_state=42)\n",
        "\n",
        "\n",
        "# Create data generator\n",
        "train_datagen = ImageDataGenerator(rescale=1./255,\n",
        "                                  horizontal_flip=True,\n",
        "                                  rotation_range=45,\n",
        "                                  brightness_range=[0.2,1.0],\n",
        "                                  zoom_range=[0.5,1.0],\n",
        "                                  width_shift_range=0.1,\n",
        "                                  height_shift_range=0.1)\n",
        "test_datagen = ImageDataGenerator(rescale=1./255,\n",
        "                                  horizontal_flip=True,\n",
        "                                  rotation_range=45,\n",
        "                                  brightness_range=[0.2,1.0],\n",
        "                                  zoom_range=[0.5,1.0],)\n",
        "\n",
        "def init_model():\n",
        "    # initialize alexNet model\n",
        "    model = Sequential()\n",
        "    # Convolutional layers\n",
        "    model.add(Conv2D(kernel_size=(11, 11),\n",
        "                    activation='relu',\n",
        "                    strides=(4, 4),\n",
        "                    filters=96,\n",
        "                    padding='valid',\n",
        "                    input_shape=(227,227,3)))\n",
        "    model.add(MaxPooling2D(pool_size=(2, 2),\n",
        "                        strides=(2, 2),\n",
        "                        padding='valid'))\n",
        "    model.add(BatchNormalization())\n",
        "    model.add(Conv2D(filters=256,\n",
        "                    kernel_size=(5, 5),\n",
        "                    strides=(1, 1),\n",
        "                    padding='same',\n",
        "                    activation='relu'))\n",
        "    model.add(MaxPooling2D(pool_size=(2, 2),\n",
        "                        strides=(2, 2),\n",
        "                        padding='valid'))\n",
        "    model.add(BatchNormalization())\n",
        "    model.add(Conv2D(filters=384,\n",
        "                    kernel_size=(3, 3),\n",
        "                    strides=(1, 1),\n",
        "                    padding='same',\n",
        "                    activation='relu'))\n",
        "    model.add(Conv2D(filters=384,\n",
        "                    kernel_size=(3, 3),\n",
        "                    strides=(1, 1),\n",
        "                    padding='same',\n",
        "                    activation='relu'))\n",
        "    model.add(Conv2D(filters=256,\n",
        "                    kernel_size=(3, 3),\n",
        "                    strides=(1, 1),\n",
        "                    padding='same',\n",
        "                    activation='relu'))\n",
        "    model.add(MaxPooling2D(pool_size=(2, 2),\n",
        "                        strides=(2, 2),\n",
        "                        padding='valid'))\n",
        "    model.add(Flatten())\n",
        "    # fully connected\n",
        "    model.add(Dense(4096, activation='relu'))\n",
        "    model.add(Dropout(0.5))\n",
        "    model.add(Dense(4096, activation='relu'))\n",
        "    model.add(Dropout(0.5))\n",
        "    model.add(Dense(1000, activation='relu'))\n",
        "    # model.add(Dropout(0.5))\n",
        "    # output\n",
        "    model.add(Dense(1, activation='linear'))\n",
        "    model.compile(loss='mean_absolute_error',\n",
        "                  optimizer=Adam(lr=1e-4),\n",
        "                  metrics=['mse'])\n",
        "    \n",
        "    return model\n",
        "  \n",
        "def save(model, history):\n",
        "    print('Saving the model ...')\n",
        "    # save model structure\n",
        "    with open(model_path, \"w\") as f:\n",
        "        f.write(model.to_json())\n",
        "    # save the weights and history\n",
        "    model.save_weights(model_weight_path)\n",
        "    pickle_save(history, history_path)\n",
        "    print('done!')\n",
        "\n",
        "def load_model():\n",
        "    \"\"\"\n",
        "    Get model from Drive\n",
        "    \"\"\"\n",
        "    json_file = open(model_path, 'r')\n",
        "    model = json_file.read()\n",
        "    json_file.close()\n",
        "    model = model_from_json(model)\n",
        "    # load weights into new model\n",
        "    model.load_weights(model_weight_path)\n",
        "    return model\n",
        "\n",
        "def plot_history(H, save=True):\n",
        "    plt.figure(figsize=FIG_SIZE)\n",
        "    plt.plot(H.history['loss'])\n",
        "    plt.plot(H.history['val_loss'])\n",
        "    plt.title('model loss')\n",
        "    plt.ylabel('loss')\n",
        "    plt.xlabel('epoch')\n",
        "    plt.legend(['train', 'test'], loc='upper left')\n",
        "    plt.savefig(DIR + 'hist.png')\n",
        "    plt.show()\n",
        "    \n",
        "\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Mq2do-Ppf3ji",
        "colab_type": "code",
        "outputId": "ce8eed21-768e-4910-933e-e6f51e3f5754",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "model = init_model()\n",
        "print(model.summary())\n",
        "BATCH_SIZE = 16\n",
        "EPOCH = 100\n",
        "print('Train ', len(trainX), ' samples', 'Test ', len(testX), ' samples')\n",
        "H = model.fit_generator(\n",
        "            train_datagen.flow(trainX, trainY, batch_size=BATCH_SIZE),\n",
        "            steps_per_epoch=len(trainX) // BATCH_SIZE,\n",
        "            epochs=EPOCH,\n",
        "            validation_data=test_datagen.flow(valX, valY)\n",
        "            )\n",
        "\n",
        "model.evaluate(testX, testY)\n",
        "plot_history(H)\n",
        "save(model, H)\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_2\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_6 (Conv2D)            (None, 55, 55, 96)        34944     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_4 (MaxPooling2 (None, 27, 27, 96)        0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_3 (Batch (None, 27, 27, 96)        384       \n",
            "_________________________________________________________________\n",
            "conv2d_7 (Conv2D)            (None, 27, 27, 256)       614656    \n",
            "_________________________________________________________________\n",
            "max_pooling2d_5 (MaxPooling2 (None, 13, 13, 256)       0         \n",
            "_________________________________________________________________\n",
            "batch_normalization_4 (Batch (None, 13, 13, 256)       1024      \n",
            "_________________________________________________________________\n",
            "conv2d_8 (Conv2D)            (None, 13, 13, 384)       885120    \n",
            "_________________________________________________________________\n",
            "conv2d_9 (Conv2D)            (None, 13, 13, 384)       1327488   \n",
            "_________________________________________________________________\n",
            "conv2d_10 (Conv2D)           (None, 13, 13, 256)       884992    \n",
            "_________________________________________________________________\n",
            "max_pooling2d_6 (MaxPooling2 (None, 6, 6, 256)         0         \n",
            "_________________________________________________________________\n",
            "flatten_2 (Flatten)          (None, 9216)              0         \n",
            "_________________________________________________________________\n",
            "dense_5 (Dense)              (None, 4096)              37752832  \n",
            "_________________________________________________________________\n",
            "dropout_4 (Dropout)          (None, 4096)              0         \n",
            "_________________________________________________________________\n",
            "dense_6 (Dense)              (None, 4096)              16781312  \n",
            "_________________________________________________________________\n",
            "dropout_5 (Dropout)          (None, 4096)              0         \n",
            "_________________________________________________________________\n",
            "dense_7 (Dense)              (None, 1000)              4097000   \n",
            "_________________________________________________________________\n",
            "dense_8 (Dense)              (None, 1)                 1001      \n",
            "=================================================================\n",
            "Total params: 62,380,753\n",
            "Trainable params: 62,380,049\n",
            "Non-trainable params: 704\n",
            "_________________________________________________________________\n",
            "None\n",
            "Train  3973  samples Test  702  samples\n",
            "Epoch 1/100\n",
            "248/248 [==============================] - 48s 193ms/step - loss: 0.7339 - mean_squared_error: 0.9147 - val_loss: 0.7693 - val_mean_squared_error: 0.8802\n",
            "Epoch 2/100\n",
            "248/248 [==============================] - 46s 187ms/step - loss: 0.6355 - mean_squared_error: 0.6369 - val_loss: 0.5624 - val_mean_squared_error: 0.4923\n",
            "Epoch 3/100\n",
            "248/248 [==============================] - 46s 186ms/step - loss: 0.5926 - mean_squared_error: 0.5631 - val_loss: 0.4931 - val_mean_squared_error: 0.3972\n",
            "Epoch 4/100\n",
            "248/248 [==============================] - 46s 185ms/step - loss: 0.5601 - mean_squared_error: 0.4939 - val_loss: 0.5988 - val_mean_squared_error: 0.5578\n",
            "Epoch 5/100\n",
            "248/248 [==============================] - 46s 185ms/step - loss: 0.5481 - mean_squared_error: 0.4821 - val_loss: 0.4638 - val_mean_squared_error: 0.3428\n",
            "Epoch 6/100\n",
            "248/248 [==============================] - 46s 184ms/step - loss: 0.5380 - mean_squared_error: 0.4611 - val_loss: 0.5589 - val_mean_squared_error: 0.5015\n",
            "Epoch 7/100\n",
            "248/248 [==============================] - 44s 179ms/step - loss: 0.5075 - mean_squared_error: 0.4133 - val_loss: 0.4562 - val_mean_squared_error: 0.3276\n",
            "Epoch 8/100\n",
            "248/248 [==============================] - 46s 187ms/step - loss: 0.5166 - mean_squared_error: 0.4255 - val_loss: 0.8185 - val_mean_squared_error: 0.9462\n",
            "Epoch 9/100\n",
            "248/248 [==============================] - 46s 185ms/step - loss: 0.4910 - mean_squared_error: 0.3947 - val_loss: 0.4532 - val_mean_squared_error: 0.3383\n",
            "Epoch 10/100\n",
            "248/248 [==============================] - 46s 186ms/step - loss: 0.4798 - mean_squared_error: 0.3741 - val_loss: 0.4750 - val_mean_squared_error: 0.3593\n",
            "Epoch 11/100\n",
            "248/248 [==============================] - 43s 175ms/step - loss: 0.4714 - mean_squared_error: 0.3584 - val_loss: 0.4900 - val_mean_squared_error: 0.3880\n",
            "Epoch 12/100\n",
            "248/248 [==============================] - 43s 175ms/step - loss: 0.4719 - mean_squared_error: 0.3651 - val_loss: 0.5437 - val_mean_squared_error: 0.4751\n",
            "Epoch 13/100\n",
            "248/248 [==============================] - 43s 173ms/step - loss: 0.4642 - mean_squared_error: 0.3549 - val_loss: 0.4271 - val_mean_squared_error: 0.3159\n",
            "Epoch 14/100\n",
            "248/248 [==============================] - 43s 172ms/step - loss: 0.4698 - mean_squared_error: 0.3595 - val_loss: 0.4142 - val_mean_squared_error: 0.2775\n",
            "Epoch 15/100\n",
            "248/248 [==============================] - 43s 173ms/step - loss: 0.4616 - mean_squared_error: 0.3463 - val_loss: 0.5068 - val_mean_squared_error: 0.4160\n",
            "Epoch 16/100\n",
            "248/248 [==============================] - 43s 173ms/step - loss: 0.4660 - mean_squared_error: 0.3522 - val_loss: 0.4663 - val_mean_squared_error: 0.3637\n",
            "Epoch 17/100\n",
            "248/248 [==============================] - 44s 176ms/step - loss: 0.4668 - mean_squared_error: 0.3507 - val_loss: 0.4938 - val_mean_squared_error: 0.3875\n",
            "Epoch 18/100\n",
            "248/248 [==============================] - 44s 179ms/step - loss: 0.4661 - mean_squared_error: 0.3564 - val_loss: 0.4212 - val_mean_squared_error: 0.2960\n",
            "Epoch 19/100\n",
            "248/248 [==============================] - 43s 171ms/step - loss: 0.4433 - mean_squared_error: 0.3200 - val_loss: 0.4390 - val_mean_squared_error: 0.3194\n",
            "Epoch 20/100\n",
            "248/248 [==============================] - 43s 172ms/step - loss: 0.4420 - mean_squared_error: 0.3187 - val_loss: 0.4980 - val_mean_squared_error: 0.3946\n",
            "Epoch 21/100\n",
            "248/248 [==============================] - 43s 173ms/step - loss: 0.4497 - mean_squared_error: 0.3280 - val_loss: 0.4133 - val_mean_squared_error: 0.2975\n",
            "Epoch 22/100\n",
            "248/248 [==============================] - 42s 171ms/step - loss: 0.4443 - mean_squared_error: 0.3285 - val_loss: 0.4834 - val_mean_squared_error: 0.3926\n",
            "Epoch 23/100\n",
            "248/248 [==============================] - 42s 171ms/step - loss: 0.4367 - mean_squared_error: 0.3144 - val_loss: 0.4226 - val_mean_squared_error: 0.3038\n",
            "Epoch 24/100\n",
            "248/248 [==============================] - 43s 172ms/step - loss: 0.4312 - mean_squared_error: 0.3030 - val_loss: 0.4038 - val_mean_squared_error: 0.2723\n",
            "Epoch 25/100\n",
            "248/248 [==============================] - 44s 176ms/step - loss: 0.4286 - mean_squared_error: 0.3007 - val_loss: 0.3968 - val_mean_squared_error: 0.2758\n",
            "Epoch 26/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.4394 - mean_squared_error: 0.3180 - val_loss: 0.4118 - val_mean_squared_error: 0.2780\n",
            "Epoch 27/100\n",
            "248/248 [==============================] - 43s 172ms/step - loss: 0.4314 - mean_squared_error: 0.3044 - val_loss: 0.4532 - val_mean_squared_error: 0.3464\n",
            "Epoch 28/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.4266 - mean_squared_error: 0.2998 - val_loss: 0.4335 - val_mean_squared_error: 0.3180\n",
            "Epoch 29/100\n",
            "248/248 [==============================] - 43s 172ms/step - loss: 0.4242 - mean_squared_error: 0.2976 - val_loss: 0.4360 - val_mean_squared_error: 0.3172\n",
            "Epoch 30/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.4193 - mean_squared_error: 0.2926 - val_loss: 0.4044 - val_mean_squared_error: 0.2719\n",
            "Epoch 31/100\n",
            "248/248 [==============================] - 42s 171ms/step - loss: 0.4185 - mean_squared_error: 0.2884 - val_loss: 0.4017 - val_mean_squared_error: 0.2744\n",
            "Epoch 32/100\n",
            "248/248 [==============================] - 43s 173ms/step - loss: 0.4213 - mean_squared_error: 0.2906 - val_loss: 0.4764 - val_mean_squared_error: 0.3872\n",
            "Epoch 33/100\n",
            "248/248 [==============================] - 42s 171ms/step - loss: 0.4086 - mean_squared_error: 0.2788 - val_loss: 0.4289 - val_mean_squared_error: 0.3140\n",
            "Epoch 34/100\n",
            "248/248 [==============================] - 43s 172ms/step - loss: 0.4115 - mean_squared_error: 0.2753 - val_loss: 0.3886 - val_mean_squared_error: 0.2532\n",
            "Epoch 35/100\n",
            "248/248 [==============================] - 42s 169ms/step - loss: 0.4161 - mean_squared_error: 0.2805 - val_loss: 0.3881 - val_mean_squared_error: 0.2486\n",
            "Epoch 36/100\n",
            "248/248 [==============================] - 43s 174ms/step - loss: 0.4101 - mean_squared_error: 0.2758 - val_loss: 0.3929 - val_mean_squared_error: 0.2654\n",
            "Epoch 37/100\n",
            "248/248 [==============================] - 43s 173ms/step - loss: 0.4089 - mean_squared_error: 0.2759 - val_loss: 0.4421 - val_mean_squared_error: 0.3290\n",
            "Epoch 38/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.4133 - mean_squared_error: 0.2815 - val_loss: 0.4102 - val_mean_squared_error: 0.2821\n",
            "Epoch 39/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.4086 - mean_squared_error: 0.2729 - val_loss: 0.4559 - val_mean_squared_error: 0.3305\n",
            "Epoch 40/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.4028 - mean_squared_error: 0.2669 - val_loss: 0.3679 - val_mean_squared_error: 0.2352\n",
            "Epoch 41/100\n",
            "248/248 [==============================] - 44s 176ms/step - loss: 0.4070 - mean_squared_error: 0.2753 - val_loss: 0.4389 - val_mean_squared_error: 0.3257\n",
            "Epoch 42/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.4029 - mean_squared_error: 0.2686 - val_loss: 0.3717 - val_mean_squared_error: 0.2382\n",
            "Epoch 43/100\n",
            "248/248 [==============================] - 42s 169ms/step - loss: 0.4085 - mean_squared_error: 0.2778 - val_loss: 0.3965 - val_mean_squared_error: 0.2709\n",
            "Epoch 44/100\n",
            "248/248 [==============================] - 43s 175ms/step - loss: 0.4019 - mean_squared_error: 0.2668 - val_loss: 0.3615 - val_mean_squared_error: 0.2189\n",
            "Epoch 45/100\n",
            "248/248 [==============================] - 43s 173ms/step - loss: 0.3919 - mean_squared_error: 0.2574 - val_loss: 0.4261 - val_mean_squared_error: 0.3058\n",
            "Epoch 46/100\n",
            "248/248 [==============================] - 42s 171ms/step - loss: 0.4050 - mean_squared_error: 0.2692 - val_loss: 0.3989 - val_mean_squared_error: 0.2665\n",
            "Epoch 47/100\n",
            "248/248 [==============================] - 42s 171ms/step - loss: 0.3955 - mean_squared_error: 0.2621 - val_loss: 0.3771 - val_mean_squared_error: 0.2415\n",
            "Epoch 48/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.4001 - mean_squared_error: 0.2659 - val_loss: 0.3796 - val_mean_squared_error: 0.2479\n",
            "Epoch 49/100\n",
            "248/248 [==============================] - 42s 171ms/step - loss: 0.3956 - mean_squared_error: 0.2614 - val_loss: 0.3707 - val_mean_squared_error: 0.2329\n",
            "Epoch 50/100\n",
            "248/248 [==============================] - 45s 182ms/step - loss: 0.3853 - mean_squared_error: 0.2485 - val_loss: 0.3889 - val_mean_squared_error: 0.2631\n",
            "Epoch 51/100\n",
            "248/248 [==============================] - 42s 169ms/step - loss: 0.3799 - mean_squared_error: 0.2404 - val_loss: 0.3687 - val_mean_squared_error: 0.2431\n",
            "Epoch 52/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.3875 - mean_squared_error: 0.2488 - val_loss: 0.4354 - val_mean_squared_error: 0.3196\n",
            "Epoch 53/100\n",
            "248/248 [==============================] - 42s 171ms/step - loss: 0.4015 - mean_squared_error: 0.2638 - val_loss: 0.3825 - val_mean_squared_error: 0.2448\n",
            "Epoch 54/100\n",
            "248/248 [==============================] - 43s 173ms/step - loss: 0.3853 - mean_squared_error: 0.2469 - val_loss: 0.3665 - val_mean_squared_error: 0.2285\n",
            "Epoch 55/100\n",
            "248/248 [==============================] - 42s 171ms/step - loss: 0.3884 - mean_squared_error: 0.2504 - val_loss: 0.4114 - val_mean_squared_error: 0.2778\n",
            "Epoch 56/100\n",
            "248/248 [==============================] - 43s 175ms/step - loss: 0.3939 - mean_squared_error: 0.2560 - val_loss: 0.3992 - val_mean_squared_error: 0.2677\n",
            "Epoch 57/100\n",
            "248/248 [==============================] - 42s 171ms/step - loss: 0.3877 - mean_squared_error: 0.2507 - val_loss: 0.3854 - val_mean_squared_error: 0.2542\n",
            "Epoch 58/100\n",
            "248/248 [==============================] - 43s 173ms/step - loss: 0.3831 - mean_squared_error: 0.2439 - val_loss: 0.3509 - val_mean_squared_error: 0.2106\n",
            "Epoch 59/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.3739 - mean_squared_error: 0.2380 - val_loss: 0.4125 - val_mean_squared_error: 0.2847\n",
            "Epoch 60/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.3851 - mean_squared_error: 0.2471 - val_loss: 0.4039 - val_mean_squared_error: 0.2829\n",
            "Epoch 61/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.3838 - mean_squared_error: 0.2410 - val_loss: 0.4060 - val_mean_squared_error: 0.2829\n",
            "Epoch 62/100\n",
            "248/248 [==============================] - 43s 174ms/step - loss: 0.3754 - mean_squared_error: 0.2383 - val_loss: 0.3675 - val_mean_squared_error: 0.2297\n",
            "Epoch 63/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.3765 - mean_squared_error: 0.2331 - val_loss: 0.3504 - val_mean_squared_error: 0.2158\n",
            "Epoch 64/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.3687 - mean_squared_error: 0.2293 - val_loss: 0.3573 - val_mean_squared_error: 0.2221\n",
            "Epoch 65/100\n",
            "248/248 [==============================] - 42s 169ms/step - loss: 0.3709 - mean_squared_error: 0.2334 - val_loss: 0.3517 - val_mean_squared_error: 0.2044\n",
            "Epoch 66/100\n",
            "248/248 [==============================] - 42s 168ms/step - loss: 0.3794 - mean_squared_error: 0.2394 - val_loss: 0.3794 - val_mean_squared_error: 0.2461\n",
            "Epoch 67/100\n",
            "248/248 [==============================] - 42s 169ms/step - loss: 0.3668 - mean_squared_error: 0.2213 - val_loss: 0.3736 - val_mean_squared_error: 0.2331\n",
            "Epoch 68/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.3766 - mean_squared_error: 0.2382 - val_loss: 0.3526 - val_mean_squared_error: 0.2131\n",
            "Epoch 69/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.3692 - mean_squared_error: 0.2307 - val_loss: 0.3402 - val_mean_squared_error: 0.1930\n",
            "Epoch 70/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.3742 - mean_squared_error: 0.2331 - val_loss: 0.3560 - val_mean_squared_error: 0.2132\n",
            "Epoch 71/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.3665 - mean_squared_error: 0.2246 - val_loss: 0.3386 - val_mean_squared_error: 0.1978\n",
            "Epoch 72/100\n",
            "248/248 [==============================] - 43s 174ms/step - loss: 0.3654 - mean_squared_error: 0.2220 - val_loss: 0.3654 - val_mean_squared_error: 0.2204\n",
            "Epoch 73/100\n",
            "248/248 [==============================] - 42s 169ms/step - loss: 0.3606 - mean_squared_error: 0.2180 - val_loss: 0.3531 - val_mean_squared_error: 0.2253\n",
            "Epoch 74/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.3613 - mean_squared_error: 0.2194 - val_loss: 0.3380 - val_mean_squared_error: 0.2044\n",
            "Epoch 75/100\n",
            "248/248 [==============================] - 42s 168ms/step - loss: 0.3561 - mean_squared_error: 0.2114 - val_loss: 0.3573 - val_mean_squared_error: 0.2301\n",
            "Epoch 76/100\n",
            "248/248 [==============================] - 43s 173ms/step - loss: 0.3656 - mean_squared_error: 0.2225 - val_loss: 0.3783 - val_mean_squared_error: 0.2469\n",
            "Epoch 77/100\n",
            "248/248 [==============================] - 42s 169ms/step - loss: 0.3576 - mean_squared_error: 0.2155 - val_loss: 0.3711 - val_mean_squared_error: 0.2383\n",
            "Epoch 78/100\n",
            "248/248 [==============================] - 45s 180ms/step - loss: 0.3604 - mean_squared_error: 0.2176 - val_loss: 0.3429 - val_mean_squared_error: 0.1952\n",
            "Epoch 79/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.3561 - mean_squared_error: 0.2150 - val_loss: 0.3279 - val_mean_squared_error: 0.1896\n",
            "Epoch 80/100\n",
            "248/248 [==============================] - 45s 181ms/step - loss: 0.3622 - mean_squared_error: 0.2194 - val_loss: 0.3534 - val_mean_squared_error: 0.2173\n",
            "Epoch 81/100\n",
            "248/248 [==============================] - 42s 169ms/step - loss: 0.3541 - mean_squared_error: 0.2102 - val_loss: 0.3954 - val_mean_squared_error: 0.2638\n",
            "Epoch 82/100\n",
            "248/248 [==============================] - 42s 169ms/step - loss: 0.3465 - mean_squared_error: 0.2019 - val_loss: 0.3629 - val_mean_squared_error: 0.2286\n",
            "Epoch 83/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.3487 - mean_squared_error: 0.2094 - val_loss: 0.3495 - val_mean_squared_error: 0.2114\n",
            "Epoch 84/100\n",
            "248/248 [==============================] - 42s 169ms/step - loss: 0.3547 - mean_squared_error: 0.2128 - val_loss: 0.3265 - val_mean_squared_error: 0.1917\n",
            "Epoch 85/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.3476 - mean_squared_error: 0.2047 - val_loss: 0.3407 - val_mean_squared_error: 0.2054\n",
            "Epoch 86/100\n",
            "248/248 [==============================] - 42s 169ms/step - loss: 0.3625 - mean_squared_error: 0.2212 - val_loss: 0.3345 - val_mean_squared_error: 0.1956\n",
            "Epoch 87/100\n",
            "248/248 [==============================] - 42s 171ms/step - loss: 0.3450 - mean_squared_error: 0.2004 - val_loss: 0.3271 - val_mean_squared_error: 0.1877\n",
            "Epoch 88/100\n",
            "248/248 [==============================] - 42s 168ms/step - loss: 0.3462 - mean_squared_error: 0.2056 - val_loss: 0.3866 - val_mean_squared_error: 0.2550\n",
            "Epoch 89/100\n",
            "248/248 [==============================] - 42s 170ms/step - loss: 0.3495 - mean_squared_error: 0.2049 - val_loss: 0.3276 - val_mean_squared_error: 0.1872\n",
            "Epoch 90/100\n",
            "248/248 [==============================] - 43s 173ms/step - loss: 0.3430 - mean_squared_error: 0.1984 - val_loss: 0.3471 - val_mean_squared_error: 0.2130\n",
            "Epoch 91/100\n",
            " 50/248 [=====>........................] - ETA: 25s - loss: 0.3548 - mean_squared_error: 0.2138"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "x1j9NoghgWSo",
        "colab_type": "text"
      },
      "source": [
        "### Test *model*"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "o4LOoLO3mTXf",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "\n",
        "from google.colab.patches import cv2_imshow\n",
        "import pandas as pd\n",
        "import cv2\n",
        "import numpy as np\n",
        "import urllib.request\n",
        "from PIL import Image\n",
        "\n",
        "model = load_model()\n",
        "model.compile(loss='mean_absolute_error',\n",
        "                optimizer=Adam(lr=1e-4),\n",
        "                metrics=['mse'])\n",
        "\n",
        "\n",
        "r = model.evaluate(testX / 255.0, testY)\n",
        "print(model.metrics_names)\n",
        "print(r)\n",
        "# pre = model.predict(testX[:1] / 255.0)\n",
        "# print(pre, testY[:1])\n",
        "# H = pickle_load(history_path)\n",
        "# plot_history(H)\n",
        "\n",
        "def isvalid(img):\n",
        "    return img.shape[0] > 0 and img.shape[1] > 0\n",
        "\n",
        "def val(img):\n",
        "    return img.shape[0] * img.shape[1]\n",
        "\n",
        "def getface(img):\n",
        "    allfaces = []\n",
        "    face_cascade = cv2.CascadeClassifier( DIR + '/haarcascade_frontalface_default.xml')\n",
        "    try:\n",
        "      gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)\n",
        "      faces = face_cascade.detectMultiScale(gray, 1.1, 4)\n",
        "      height, width = img.shape[:2]\n",
        "      for (x, y, w, h) in faces:\n",
        "          face = img[y :min(y + h, height),\n",
        "                    x -50 :min(x + w, width)]\n",
        "          # only get the first one\n",
        "          if (isvalid(face)):\n",
        "            allfaces.append(face)\n",
        "    except:\n",
        "      return img\n",
        "    \n",
        "    if (len(allfaces)):\n",
        "      return sorted(allfaces, key=lambda x: -val(x))[0]\n",
        "    return img\n",
        "\n",
        "def predict(img_path):\n",
        "    img = None\n",
        "    if img_path.startswith('http') or \\\n",
        "       img_path.startswith('data:image/jpeg'):\n",
        "        req = urllib.request.urlopen(img_path)\n",
        "        arr = np.asarray(bytearray(req.read()), dtype=np.uint8)\n",
        "        img = cv2.imdecode(arr, -1)\n",
        "    else:\n",
        "        img = cv2.imread(img_path)\n",
        "    \n",
        "    # Crop the face\n",
        "    img = getface(img)\n",
        "    try:\n",
        "      img = cv2.resize(img, (227, 227))\n",
        "    except:\n",
        "      pass\n",
        "    cv2_imshow(img)  \n",
        "    img = np.expand_dims(img / 255.0, axis=0)\n",
        "    pre = model.predict(img)[0]\n",
        "    print(str(round(pre[0], 2)))\n",
        "    return pre\n",
        "\n",
        "imgpaths = [DIR + 'test_imgs/' + path for path in os.listdir(DIR + 'test_imgs')]\n",
        "\n",
        "\n",
        "for path in imgpaths:\n",
        "  predict(path)\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5WK1-j-EeRXZ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "07d19251-11a4-4948-cb84-35862dfd901a"
      },
      "source": [
        "def plot_hist_per_label(group):\n",
        "    for gr, vals in group.items():\n",
        "        plt.hist(vals, bins=40, label = gr)\n",
        "        # sns.distplot(vals, kde=True, hist=False, label=gr)\n",
        "        # plt.legend(loc='upper right')\n",
        "        plt.xlabel('Score')\n",
        "        plt.ylabel('Frequency')\n",
        "        plt.savefig(DIR + 'dataset_' + gr + '.png') \n",
        "        plt.show()\n",
        "\n",
        "\n",
        "def plot_one_density(group):\n",
        "    plt.figure(figsize=FIG_SIZE)\n",
        "    for gr, vals in group.items():\n",
        "        sns.distplot(vals, kde=True, hist=False, label=gr)\n",
        "    plt.legend( title = 'Group')\n",
        "    plt.title('Density Plot dataset distribution')\n",
        "    plt.xlabel('Score')\n",
        "    plt.ylabel('Density')\n",
        "\n",
        "def dataset_distribute():\n",
        "    df = pd.read_csv(LABELS)\n",
        "    group_ = {\n",
        "        'AF': [],\n",
        "        'AM': [],\n",
        "        'CF': [],\n",
        "        'CM': [],\n",
        "    }\n",
        "\n",
        "    def fgroup(filename):\n",
        "        for g in group_.keys():\n",
        "          if g in filename:\n",
        "            return g\n",
        "        return False\n",
        "\n",
        "    labels = df['score']\n",
        "    files = df['filename']\n",
        "    for i in range(len(labels)):\n",
        "        filename = files[i]\n",
        "        label = labels[i]\n",
        "        gname = fgroup(filename)\n",
        "        group_[gname].append(label)\n",
        "    \n",
        "    plot_hist_per_label(group_)\n",
        "    plot_one_density(group_)\n",
        "\n",
        "       \n",
        "\n",
        "dataset_distribute()\n",
        "# H = pickle_load(history_path)\n",
        "# plot_history(H)\n"
      ],
      "execution_count": 33,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD4CAYAAAAXUaZHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nktQgy12SGmS5S1KDLHdJatD/A1ehUkTtJ4bgAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD4CAYAAAAXUaZHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXAAAAD4CAYAAAD1jb0+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0\ndHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAOCUlEQVR4nO3da4ym9V3G8e9VDtJYKqU7rhuW7ZBA\nNGgEdLKlwRcVxNAuARIJ0mjdGswmxkYam9RtX2hqfAFv2nqKugHSrdYWAm1ZaatugKY2sbS7QFtg\n27CSJUKA3XImGszizxdzLwyzM/PcOzPP4T/7/SSbee7D8FzczF78538fnlQVkqT2vGXcASRJy2OB\nS1KjLHBJapQFLkmNssAlqVEnjvLN1q1bV9PT06N8S0lq3t69e39cVVPz14+0wKenp9mzZ88o31KS\nmpfk8YXWO4UiSY2ywCWpURa4JDXKApekRlngktQoC1ySGtXrMsIkB4CXgdeAw1U1k+R04FZgGjgA\nXFNVzw8npiRpvmMZgf9qVZ1fVTPd8nbg7qo6B7i7W5YkjchKplCuBHZ2r3cCV608jiSpr753Yhbw\nb0kK+Puq2gGsr6qnuu1PA+sX+sYk24BtAJs2bVphXLVuevtXl9x+4IYtI0oita9vgf9KVT2Z5KeB\n3Ul+OHdjVVVX7kfpyn4HwMzMjB//I0mrpNcUSlU92X09CHwZ2Aw8k2QDQPf14LBCSpKONrDAk/xk\nklOPvAZ+HXgI2AVs7XbbCtw5rJCSpKP1mUJZD3w5yZH9/6mq/iXJd4HbklwHPA5cM7yYOl4sNUfu\n/Lj0ZgMLvKoeA85bYP2zwCXDCCVJGsw7MSWpURa4JDXKApekRlngktQoC1ySGmWBS1KjLHBJapQF\nLkmNssAlqVEWuCQ1ygKXpEZZ4JLUKAtckhplgUtSoyxwSWqUBS5JjbLAJalRFrgkNcoCl6RGWeCS\n1CgLXJIaZYFLUqMscElqlAUuSY2ywCWpURa4JDXKApekRlngktQoC1ySGmWBS1KjLHBJapQFLkmN\n6l3gSU5I8kCSu7rls5Lcl2R/kluTnDy8mJKk+Y5lBH49sG/O8o3Ap6vqbOB54LrVDCZJWlqvAk+y\nEdgC3NQtB7gYuL3bZSdw1TACSpIWdmLP/T4DfAw4tVt+J/BCVR3ulp8AzljoG5NsA7YBbNq0aflJ\nddyb3v7VJbcfuGHLiJJIk2HgCDzJ5cDBqtq7nDeoqh1VNVNVM1NTU8v5R0iSFtBnBH4RcEWS9wOn\nAG8H/gI4LcmJ3Sh8I/Dk8GJKkuYbOAKvqo9X1caqmgauBe6pqt8C7gWu7nbbCtw5tJSSpKOs5Drw\nPwb+KMl+ZufEb16dSJKkPvqexASgqr4BfKN7/RiwefUjSZL68E5MSWqUBS5JjbLAJalRFrgkNcoC\nl6RGWeCS1CgLXJIaZYFLUqMscElqlAUuSY2ywCWpURa4JDXKApekRlngktQoC1ySGnVMzwOXWrbU\nhyL7gchqkSNwSWqUBS5JjbLAJalRzoFrzVhqjltaixyBS1KjLHBJapQFLkmNcg5cq865aGk0HIFL\nUqMscElqlAUuSY2ywCWpURa4JDXKApekRlngktQoC1ySGjWwwJOckuQ7Sb6X5OEkn+zWn5XkviT7\nk9ya5OThx5UkHdFnBP4qcHFVnQecD1yW5ELgRuDTVXU28Dxw3fBiSpLmG1jgNeuVbvGk7k8BFwO3\nd+t3AlcNJaEkaUG95sCTnJDkQeAgsBv4T+CFqjrc7fIEcMZwIkqSFtKrwKvqtao6H9gIbAZ+ru8b\nJNmWZE+SPYcOHVpmTEnSfMd0FUpVvQDcC7wHOC3JkacZbgSeXOR7dlTVTFXNTE1NrSisJOkNfa5C\nmUpyWvf6rcClwD5mi/zqbretwJ3DCilJOlqf54FvAHYmOYHZwr+tqu5K8gjwxSR/DjwA3DzEnJKk\neQYWeFV9H7hggfWPMTsfrgYt9aELB27YMsIkatGgD+3wZ2g0vBNTkhplgUtSoyxwSWqUH2qsozi/\nKbXBEbgkNcoCl6RGWeCS1CjnwKWGeT3/8c0RuCQ1ygKXpEZZ4JLUKOfAJQZf+74U55qP5r0Eo+EI\nXJIaZYFLUqMscElqlHPgOmYrmS+WtHocgUtSoyxwSWqUBS5JjXIOXFqjvBZ77XMELkmNssAlqVEW\nuCQ1ygKXpEZ5ErNhPsxferPj7cStI3BJapQFLkmNssAlqVEWuCQ1ygKXpEZZ4JLUKAtckhplgUtS\nowYWeJIzk9yb5JEkDye5vlt/epLdSR7tvr5j+HElSUf0GYEfBj5aVecCFwJ/kORcYDtwd1WdA9zd\nLUuSRmRggVfVU1V1f/f6ZWAfcAZwJbCz220ncNWwQkqSjnZMz0JJMg1cANwHrK+qp7pNTwPrF/me\nbcA2gE2bNi03p6RVdrw9N2Qt6n0SM8nbgDuAj1TVS3O3VVUBtdD3VdWOqpqpqpmpqakVhZUkvaFX\ngSc5idny/nxVfalb/UySDd32DcDB4USUJC2kz1UoAW4G9lXVp+Zs2gVs7V5vBe5c/XiSpMX0mQO/\nCPgg8IMkD3brPgHcANyW5DrgceCa4USUdDxxbr6/gQVeVd8CssjmS1Y3jiSpL+/ElKRGWeCS1CgL\nXJIaZYFLUqMscElqlAUuSY06pmehSDp2Xtc8OdbafwtH4JLUKAtckhplgUtSoyxwSWqUJzGlMRt0\nYk1ajCNwSWqUBS5JjbLAJalRzoGvUc6raqX8GZp8jsAlqVEWuCQ1ygKXpEY5By5JnaXm/SfxQVeO\nwCWpURa4JDXKApekRjkHLqkpXp/+BkfgktQoC1ySGmWBS1KjnAOXNHLOY68OR+CS1CgLXJIaZYFL\nUqOcA59gzhO2wf9OGpeBI/AktyQ5mOShOetOT7I7yaPd13cMN6Ykab4+UyifBS6bt247cHdVnQPc\n3S1LkkZoYIFX1TeB5+atvhLY2b3eCVy1yrkkSQMs9yTm+qp6qnv9NLB+sR2TbEuyJ8meQ4cOLfPt\nJEnzrfgqlKoqoJbYvqOqZqpqZmpqaqVvJ0nqLLfAn0myAaD7enD1IkmS+lhuge8CtnavtwJ3rk4c\nSVJffS4j/ALwH8DPJnkiyXXADcClSR4Ffq1bliSN0MAbearqA4tsumSVs0iSjoG30ktSoyxwSWqU\nBS5JjfJhVkO21IOODtywZYRJJK01jsAlqVEWuCQ1ygKXpEZZ4JLUKAtckhplgUtSoyxwSWqUBS5J\njbLAJalRFrgkNcoCl6RG+SyUMVrqOSmSJsugv6/jeLaRI3BJapQFLkmNssAlqVHOgUvSKhjHs/8d\ngUtSoyxwSWqUBS5JjbLAJalRFrgkNcoCl6RGWeCS1CgLXJIa5Y08A0ziA2wkCRyBS1KzLHBJapQF\nLkmNamYOfJhz0Sv5YAU/lEHSuKxoBJ7ksiQ/SrI/yfbVCiVJGmzZBZ7kBOBvgPcB5wIfSHLuagWT\nJC1tJSPwzcD+qnqsqv4X+CJw5erEkiQNspI58DOA/5qz/ATw7vk7JdkGbOsWX0nyoxW856Jy45Kb\n1wE/Hsb7DlGLmaHN3GYejeM284B+6uNdC60c+knMqtoB7Bj2+ywlyZ6qmhlnhmPVYmZoM7eZR8PM\nq28lUyhPAmfOWd7YrZMkjcBKCvy7wDlJzkpyMnAtsGt1YkmSBln2FEpVHU7yYeBfgROAW6rq4VVL\ntrrGOoWzTC1mhjZzm3k0zLzKUlXjziBJWgZvpZekRlngktSoNVPgSW5JcjDJQ4tsf2+SF5M82P35\nk1FnXCDTmUnuTfJIkoeTXL/APknyl93jCr6f5JfGkXVOnj6ZJ/FYn5LkO0m+1+X+5AL7/ESSW7tj\nfV+S6dEnfVOePpk/lOTQnGP9e+PIOl+SE5I8kOSuBbZN1HE+YkDmiTzOzTzMqofPAn8NfG6Jff69\nqi4fTZxeDgMfrar7k5wK7E2yu6oembPP+4Bzuj/vBv6WBW6YGqE+mWHyjvWrwMVV9UqSk4BvJfl6\nVX17zj7XAc9X1dlJrgVuBH5zHGE7fTID3FpVHx5DvqVcD+wD3r7Atkk7zkcslRkm8DivmRF4VX0T\neG7cOY5FVT1VVfd3r19m9ofnjHm7XQl8rmZ9GzgtyYYRR31dz8wTpzt+r3SLJ3V/5p/BvxLY2b2+\nHbgkSUYU8Sg9M0+cJBuBLcBNi+wyUccZemWeSGumwHt6T/fr6NeT/Py4w8zV/Rp5AXDfvE0LPbJg\nIgpzicwwgce6+xX5QeAgsLuqFj3WVXUYeBF452hTvlmPzAC/0U2v3Z7kzAW2j9pngI8B/7fI9ok7\nzgzODJN3nI+rAr8feFdVnQf8FfCVMed5XZK3AXcAH6mql8adp48BmSfyWFfVa1V1PrN3DW9O8gvj\nzjRIj8z/DExX1S8Cu3ljZDsWSS4HDlbV3nHmOBY9M0/UcT7iuCnwqnrpyK+jVfU14KQk68Yci25u\n8w7g81X1pQV2mbhHFgzKPKnH+oiqegG4F7hs3qbXj3WSE4GfAp4dbbqFLZa5qp6tqle7xZuAXx51\ntnkuAq5IcoDZJ5RenOQf5+0zacd5YOYJPM7AcVTgSX7myDxbks3M/ruP9S9nl+dmYF9VfWqR3XYB\nv9NdjXIh8GJVPTWykPP0yTyhx3oqyWnd67cClwI/nLfbLmBr9/pq4J4a451ufTLPOx9yBbPnJMam\nqj5eVRuraprZx2vcU1W/PW+3iTrOfTJP2nE+Ys1chZLkC8B7gXVJngD+lNmTPlTV3zH7g/L7SQ4D\n/wNcO84fms5FwAeBH3TznACfADbB67m/Brwf2A/8N/C7Y8g5V5/Mk3isNwA7M/tBJG8Bbququ5L8\nGbCnqnYx+z+mf0iyn9kT4teOLy7QL/MfJrmC2auDngM+NLa0S5jw47ygFo6zt9JLUqOOmykUSVpr\nLHBJapQFLkmNssAlqVEWuCQ1ygKXpEZZ4JLUqP8HXLAiTZjGhcoAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXAAAAD4CAYAAAD1jb0+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmEAAAG5CAYAAADGcOOUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrkJggg==\n",
            "text/plain": [
              "<Figure size 720x504 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}