App_train.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.preprocessing.image import ImageDataGenerator, array_to_img, img_to_array, load_img\n",
    "from keras.models import Sequential \n",
    "from keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dropout\n",
    "from keras.layers.normalization import BatchNormalization\n",
    "from keras.callbacks import ModelCheckpoint, EarlyStopping, TensorBoard\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from keras.utils import np_utils\n",
    "from keras import optimizers\n",
    "\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.metrics import classification_report, confusion_matrix"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Facial Expression Recognition"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 0=Angry, 1=Disgust, 2=Fear, 3=Happy, 4=Sad, 5=Surprise, 6=Neutral"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def pre_pro():\n",
    "    \n",
    "    \n",
    "    data1 = pd.read_csv('./fer2013.csv')\n",
    "    \n",
    "    # Removendo as expressoes Angry=0 e Disgust=1\n",
    "    data1  = data1[data1['emotion'] != 1]\n",
    "    data1  = data1[data1['emotion'] != 0]\n",
    "    \n",
    "     \n",
    "    # Alterando os labels das expressoes Surprise de 5 para 0 e Neutral de 6 para 1 \n",
    "    data1['emotion'] = data1['emotion'].replace(5, 0) \n",
    "    data1['emotion'] = data1['emotion'].replace(6, 1)\n",
    "    \n",
    "    \n",
    "    return data1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def pre_pixels(data):\n",
    "    # Script para preparar conjuntos de dados, carregar dados e gerar imagens em escala \n",
    "    # Crie um diretório e mantenha o arquivo fer2013.csv no diretório. \n",
    "    # Dataset original em: https://www.kaggle.com/c/challenges-in-representation-learning-facial-expression-recognition-challenge\n",
    "   \n",
    "    data1 = data\n",
    "    \n",
    "    \"\"\" \n",
    "    # Removendo as expressoes Angry=0 e Disgust=1\n",
    "    data1  = data1[data1['emotion'] != 1]\n",
    "    data1  = data1[data1['emotion'] != 0]\n",
    "    \n",
    "    # Alterando os labels das expressoes Surprise de 5 para 0 e Neutral de 6 para 1 \n",
    "    data1['emotion'] = data1['emotion'].replace(5, 0) \n",
    "    data1['emotion'] = data1['emotion'].replace(6, 1) \n",
    "    \"\"\"\n",
    "\n",
    "\n",
    "    pixels1 = data1['pixels']\n",
    "\n",
    "    pix = list(pixels1)\n",
    "    pix1 = pd.DataFrame(pix)\n",
    "    pix2 = pix1.iloc[:,0].str.split(' ', expand=True)\n",
    "    df = pix2\n",
    "    \n",
    "  \n",
    "    # Normalização\n",
    "    scaler = MinMaxScaler()\n",
    "    df_n = scaler.fit_transform(df)\n",
    "    \n",
    "    # Salva os dados em formato numpy\n",
    "    np.save('data/normalize.bin.npy',df_n)\n",
    "    \n",
    "    print(\"Shape pixels\")\n",
    "    print(df_n.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def pre_emotions (data):\n",
    "    \n",
    "    \n",
    "    data1 = data['emotion']\n",
    "\n",
    "\n",
    "    \"\"\"# Removendo as expressoes Angry=0 e Disgust=1\n",
    "    data1  = data1[data1['emotion'] != 1]\n",
    "    data1  = data1[data1['emotion'] != 0]\n",
    "\n",
    "    # Alterando os labels das expressoes Surprise de 5 para 0 e Neutral de 6 para 1 \n",
    "    data1['emotion'] = data1['emotion'].replace(5, 0) \n",
    "    data1['emotion'] = data1['emotion'].replace(6, 1) \"\"\"\n",
    "    \n",
    "    data = np.array(data1)\n",
    "    data = data.astype('float64')\n",
    "\n",
    "  \n",
    "\n",
    "    np.save('./emotion.bin.npy',data)\n",
    "    \n",
    "    print(\"Shape emotion\")\n",
    "    print(data.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "############ Caso queira executar o pré-processamento novamente descomente as lihna abaixo (pode demorar um pouco para executar)\n",
    "\n",
    "\n",
    "#data = pre_pro()\n",
    "#print(\"pre_pro ok\")\n",
    "#pre_pixels(data)\n",
    "#print(\"pre_pix ok\")\n",
    "#pre_emotions(data)\n",
    "#print(\"pre_emo ok\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Carregando dados"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = np.load('./data/normalize.bin.npy')\n",
    "labels = np.load('./emotion.bin.npy')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(30387, 2304)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(30387,)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labels.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# one hot encoding\n",
    "labels = pd.DataFrame(labels)\n",
    "labels = np_utils.to_categorical(labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Dividindo dados de treinamento, validadação e teste"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "30387"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num = len(X_train)\n",
    "train = int ( num  * 0.7)\n",
    "val = int ( num  * 0.8)\n",
    "num"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dados de trainamento: 21270, val: 3039 e teste: 6078\n"
     ]
    }
   ],
   "source": [
    "X_trainn = X_train[:train]\n",
    "X_val = X_train[train:val]\n",
    "X_test = X_train[val:]\n",
    "\n",
    "total = (len(X_trainn)+  len(X_val)+ len(X_test))\n",
    "\n",
    "print(\"Dados de trainamento: %s, val: %s e teste: %s\"%(len(X_trainn), len(X_val), len(X_test)))\n",
    "total\n",
    "\n",
    "\n",
    "X_trainn = X_trainn.reshape(X_trainn.shape[0] ,  48 , 48,1)\n",
    "X_val = X_val.reshape(X_val.shape[0] ,  48 , 48,1)\n",
    "X_test = X_test.reshape(X_test.shape[0] ,  48 , 48,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "lables: 21270, val: 3039 e teste: 6078\n"
     ]
    }
   ],
   "source": [
    "y_train = labels[:train]\n",
    "y_val = labels[train:val]\n",
    "y_test = labels[val:]\n",
    "\n",
    "total = (len(y_train)+  len(y_val)+ len(y_test))\n",
    "total\n",
    "\n",
    "print(\"lables: %s, val: %s e teste: %s\"%(len(y_train), len(y_val), len(y_test)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Exibindo imagem "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "X_train = X_train.reshape(X_train.shape[0] ,  48 , 48)\n",
    "#X_train2 = X_train.reshape(X_train.shape[0] ,  48 , 48, 1)\n",
    "\n",
    "plt.imshow(X_train[0], cmap=\"gray\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Modelo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_4 (Conv2D)            (None, 46, 46, 128)       1280      \n",
      "_________________________________________________________________\n",
      "batch_normalization_4 (Batch (None, 46, 46, 128)       512       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_4 (MaxPooling2 (None, 23, 23, 128)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_5 (Conv2D)            (None, 23, 23, 64)        73792     \n",
      "_________________________________________________________________\n",
      "batch_normalization_5 (Batch (None, 23, 23, 64)        256       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_5 (MaxPooling2 (None, 11, 11, 64)        0         \n",
      "_________________________________________________________________\n",
      "dropout_5 (Dropout)          (None, 11, 11, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_6 (Conv2D)            (None, 11, 11, 32)        18464     \n",
      "_________________________________________________________________\n",
      "batch_normalization_6 (Batch (None, 11, 11, 32)        128       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_6 (MaxPooling2 (None, 5, 5, 32)          0         \n",
      "_________________________________________________________________\n",
      "dropout_6 (Dropout)          (None, 5, 5, 32)          0         \n",
      "_________________________________________________________________\n",
      "flatten_2 (Flatten)          (None, 800)               0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 512)               410112    \n",
      "_________________________________________________________________\n",
      "dropout_7 (Dropout)          (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 256)               131328    \n",
      "_________________________________________________________________\n",
      "dropout_8 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 5)                 1285      \n",
      "=================================================================\n",
      "Total params: 637,157\n",
      "Trainable params: 636,709\n",
      "Non-trainable params: 448\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "\n",
    "model.add(Conv2D(128, (3,3), input_shape = (48,48,1),padding=\"valid\", activation='relu'))\n",
    "model.add(BatchNormalization())\n",
    "model.add(MaxPooling2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Conv2D(64, (3,3), activation='relu', padding=\"same\"))\n",
    "model.add(BatchNormalization())\n",
    "model.add(MaxPooling2D(pool_size=(2,2)))\n",
    "model.add(Dropout(0.2))\n",
    "\n",
    "model.add(Conv2D(32, (3,3), activation='relu',padding=\"same\"))\n",
    "model.add(BatchNormalization())\n",
    "model.add(MaxPooling2D(pool_size=(2,2)))\n",
    "model.add(Dropout(0.2))\n",
    "\n",
    "\n",
    "\n",
    "model.add(Flatten())\n",
    "\n",
    "model.add(Dense(512, activation='relu'))\n",
    "model.add(Dropout(0.3))\n",
    "\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dropout(0.2))\n",
    "\n",
    "model.add(Dense(5, activation = 'softmax'))\n",
    "\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "opt = optimizers.Adam(lr=0.001);\n",
    "#optm = optimizers.RMSprop(lr=0.004, rho=0.9, epsilon=1e-08, decay=0.0)\n",
    "#sgd = optimizers.SGD(lr=0.055, decay=1e-6, momentum=0.9, nesterov=True)\n",
    "model.compile(optimizer=opt, loss='categorical_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 21270 samples, validate on 3039 samples\n",
      "Epoch 1/50\n",
      "21270/21270 [==============================] - 41s 2ms/step - loss: 0.4964 - acc: 0.8124 - val_loss: 0.9254 - val_acc: 0.6578\n",
      "\n",
      "Epoch 00001: val_loss improved from inf to 0.92539, saving model to weights_203.hdf5\n",
      "Epoch 2/50\n",
      "21270/21270 [==============================] - 41s 2ms/step - loss: 0.4827 - acc: 0.8142 - val_loss: 0.9515 - val_acc: 0.6660\n",
      "\n",
      "Epoch 00002: val_loss did not improve from 0.92539\n",
      "Epoch 3/50\n",
      "21270/21270 [==============================] - 41s 2ms/step - loss: 0.4732 - acc: 0.8204 - val_loss: 1.0052 - val_acc: 0.6561\n",
      "\n",
      "Epoch 00003: val_loss did not improve from 0.92539\n",
      "Epoch 4/50\n",
      "21270/21270 [==============================] - 42s 2ms/step - loss: 0.4532 - acc: 0.8272 - val_loss: 0.9510 - val_acc: 0.6598\n",
      "\n",
      "Epoch 00004: val_loss did not improve from 0.92539\n",
      "Epoch 5/50\n",
      "21270/21270 [==============================] - 42s 2ms/step - loss: 0.4455 - acc: 0.8311 - val_loss: 0.9572 - val_acc: 0.6578\n",
      "\n",
      "Epoch 00005: val_loss did not improve from 0.92539\n",
      "Epoch 6/50\n",
      "21270/21270 [==============================] - 42s 2ms/step - loss: 0.4337 - acc: 0.8379 - val_loss: 0.9190 - val_acc: 0.6578\n",
      "\n",
      "Epoch 00006: val_loss improved from 0.92539 to 0.91903, saving model to weights_203.hdf5\n",
      "Epoch 7/50\n",
      "21270/21270 [==============================] - 41s 2ms/step - loss: 0.4261 - acc: 0.8394 - val_loss: 0.9785 - val_acc: 0.6598\n",
      "\n",
      "Epoch 00007: val_loss did not improve from 0.91903\n",
      "Epoch 8/50\n",
      "21270/21270 [==============================] - 42s 2ms/step - loss: 0.4159 - acc: 0.8411 - val_loss: 0.9482 - val_acc: 0.6499\n",
      "\n",
      "Epoch 00008: val_loss did not improve from 0.91903\n",
      "Epoch 9/50\n",
      "21270/21270 [==============================] - 42s 2ms/step - loss: 0.3963 - acc: 0.8522 - val_loss: 1.0099 - val_acc: 0.6637\n",
      "\n",
      "Epoch 00009: val_loss did not improve from 0.91903\n",
      "Epoch 10/50\n",
      "21270/21270 [==============================] - 41s 2ms/step - loss: 0.3977 - acc: 0.8495 - val_loss: 1.1166 - val_acc: 0.6528\n",
      "\n",
      "Epoch 00010: val_loss did not improve from 0.91903\n",
      "Epoch 11/50\n",
      "21270/21270 [==============================] - 43s 2ms/step - loss: 0.3755 - acc: 0.8603 - val_loss: 1.1130 - val_acc: 0.6594\n",
      "\n",
      "Epoch 00011: val_loss did not improve from 0.91903\n",
      "Epoch 12/50\n",
      "21270/21270 [==============================] - 43s 2ms/step - loss: 0.3772 - acc: 0.8587 - val_loss: 1.0536 - val_acc: 0.6614\n",
      "\n",
      "Epoch 00012: val_loss did not improve from 0.91903\n",
      "Epoch 13/50\n",
      "21270/21270 [==============================] - 44s 2ms/step - loss: 0.3670 - acc: 0.8619 - val_loss: 0.9795 - val_acc: 0.6637\n",
      "\n",
      "Epoch 00013: val_loss did not improve from 0.91903\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7f3fa46ae668>"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Salvando o melhor modelo (melhor conjunto de pesos)\n",
    "check = ModelCheckpoint(filepath='weights_203.hdf5', verbose=1, save_best_only=True)\n",
    "early = EarlyStopping(monitor='val_loss',patience=7)\n",
    "tensorboard = TensorBoard(log_dir='./logs')\n",
    "\n",
    "model.fit(X_trainn, y_train, batch_size=100 , epochs=50, validation_data=[X_val, y_val], callbacks=[check, early, tensorboard])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#model.load_weights('./weights.hdf5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "########## Caso Deseje salvar uma nova arquitura, descomente as linha abaixo \n",
    "#import h5py\n",
    "#json_string = model.to_json()\n",
    "#open('models/model_arch.json', 'w').write(json_string)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(6078, 5)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predictions = model.predict(X_test)\n",
    "predictions.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f3eb13674e0>"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sn\n",
    "matrix = confusion_matrix(y_test.argmax(axis=1), predictions.argmax(axis=1))\n",
    "matrix\n",
    "plt.figure(figsize = (20,10))\n",
    "sn.heatmap(matrix, annot=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "target_names = ['Surprise', 'Neutral', 'Fear', 'Happy', 'Sad' ]\n",
    "#print(classification_report(y_test, predictions, target_names=target_names))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plott():\n",
    "    normalize=False\n",
    "    title=None\n",
    "    cmap=plt.cm.Blues\n",
    "    cm = matrix\n",
    "    classes = target_names\n",
    "\n",
    "    fig, ax = plt.subplots(figsize = (20,10))\n",
    "    im = ax.imshow(matrix, interpolation='nearest', cmap=cmap)\n",
    "    ax.figure.colorbar(im, ax=ax)\n",
    "    # We want to show all ticks...\n",
    "    ax.set(xticks=np.arange(cm.shape[1]),\n",
    "           yticks=np.arange(cm.shape[0]),\n",
    "           # ... and label them with the respective list entries\n",
    "           xticklabels=classes, yticklabels=classes,\n",
    "           title=title,\n",
    "           ylabel='True label',\n",
    "           xlabel='Predicted label')\n",
    "\n",
    "    # Rotate the tick labels and set their alignment.\n",
    "    plt.setp(ax.get_xticklabels(), rotation=45, ha=\"right\",\n",
    "             rotation_mode=\"anchor\")\n",
    "\n",
    "    # Loop over data dimensions and create text annotations.\n",
    "    fmt = '.2f' if normalize else 'd'\n",
    "    thresh = matrix.max() / 2.\n",
    "    for i in range(matrix.shape[0]):\n",
    "        for j in range(matrix.shape[1]):\n",
    "            ax.text(j, i, format(matrix[i, j], fmt),\n",
    "                    ha=\"center\", va=\"center\",\n",
    "                    color=\"white\" if matrix[i, j] > thresh else \"black\")\n",
    "    fig.tight_layout()\n",
    "    return ax"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f3eb0bb1748>"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plott()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:cuda9.2]",
   "language": "python",
   "name": "conda-env-cuda9.2-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}