main.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "main.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "metadata": {
        "id": "krzbanZM4Oht",
        "colab_type": "code",
        "outputId": "e47e40e8-bd35-44ca-f508-df42add182ad",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1245
        }
      },
      "cell_type": "code",
      "source": [
        "'''\n",
        "CS 559: Deep Learning\n",
        "Homework: Age Estimation Using TensorFlow\n",
        "Muhammed Cavusoglu (21400653) and Kemal Buyukkaya (21200496)\n",
        "'''\n",
        "\n",
        "import os\n",
        "import cv2\n",
        "import numpy as np\n",
        "import tensorflow as tf\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "# For colab\n",
        "# '''\n",
        "from google.colab import drive\n",
        "drive.mount('/content/drive')\n",
        "# '''\n",
        "\n",
        "tf.logging.set_verbosity(tf.logging.INFO)\n",
        "\n",
        "def cnn_model(features, labels, mode):\n",
        "    # Input Layer\n",
        "    input_layer = tf.reshape(features[\"x\"], [-1, 88, 88, 1])\n",
        "    \n",
        "    # Conv1 Layer\n",
        "    # Compute 32 features using a 4 x 4 filter with ReLU activation\n",
        "    # Width and height values are preserved (padding)\n",
        "    # Input Tensor Shape: [batch_size, 88, 88, 1]\n",
        "    # Output Tensor Shape: [batch_size, 88, 88, 32]\n",
        "    conv1 = tf.contrib.layers.conv2d(\n",
        "        inputs = input_layer,\n",
        "        num_outputs = 32,\n",
        "        kernel_size = [4, 4],\n",
        "        padding = 'SAME',\n",
        "        activation_fn = tf.nn.relu,\n",
        "        weights_initializer = tf.contrib.layers.xavier_initializer(),\n",
        "        weights_regularizer = tf.contrib.layers.l2_regularizer(0.8)\n",
        "    )\n",
        "    \n",
        "    # Conv2 Layer\n",
        "    # Compute 32 features using a 4 x 4 filter with ReLU activation\n",
        "    # Width and height values are preserved (padding)\n",
        "    # Input Tensor Shape: [batch_size, 88, 88, 32]\n",
        "    # Output Tensor Shape: [batch_size, 88, 88, 32]\n",
        "    conv2 = tf.contrib.layers.conv2d(\n",
        "        inputs = conv1,\n",
        "        num_outputs = 32,\n",
        "        kernel_size = [4, 4],\n",
        "        padding= 'SAME',\n",
        "        activation_fn = tf.nn.relu,\n",
        "        weights_initializer = tf.contrib.layers.xavier_initializer(),\n",
        "        weights_regularizer = tf.contrib.layers.l2_regularizer(0.8)\n",
        "    )\n",
        "    \n",
        "    # Pooling1 Layer\n",
        "    # Max Pooling layer with a 2 x 2 filter and stride of 2\n",
        "    # Input Tensor Shape: [batch_size, 88, 88, 32]\n",
        "    # Output Tensor Shape: [batch_size, 44, 44, 32]\n",
        "    pool1 = tf.contrib.layers.max_pool2d(inputs = conv2, kernel_size = [2, 2], stride = 2)\n",
        "    \n",
        "    # BatchNorm Layer\n",
        "    bn1 = tf.contrib.layers.batch_norm(inputs = pool1, activation_fn = None)\n",
        "\n",
        "    # Conv3 Layer\n",
        "    # Compute 64 features using a 4 x 4 filter with ReLU activation\n",
        "    # Width and height values are preserved (padding)\n",
        "    # Input Tensor Shape: [batch_size, 44, 44, 32]\n",
        "    # Output Tensor Shape: [batch_size, 44, 44, 64]\n",
        "    conv3 = tf.contrib.layers.conv2d(\n",
        "        inputs = bn1,\n",
        "        num_outputs = 64,\n",
        "        kernel_size = [4, 4],\n",
        "        padding= 'SAME',\n",
        "        activation_fn = tf.nn.relu,\n",
        "        weights_initializer=tf.contrib.layers.xavier_initializer(),\n",
        "        weights_regularizer = tf.contrib.layers.l2_regularizer(0.8)\n",
        "    )\n",
        "    \n",
        "    # Pooling 2 Layer\n",
        "    # Max Pooling layer with a 2 x 2 filter and stride of 2\n",
        "    # Input Tensor Shape: [batch_size, 44, 44, 64]\n",
        "    # Output Tensor Shape: [batch_size, 22, 22, 64]\n",
        "    pool2 = tf.contrib.layers.max_pool2d(inputs = conv3, kernel_size = [2, 2], stride = 2)\n",
        "    \n",
        "    # BatchNorm Layer\n",
        "    bn2 = tf.contrib.layers.batch_norm(inputs = pool2, activation_fn = None)\n",
        "    \n",
        "    # Reshape Layer\n",
        "    # Input Tensor Shape: [batch_size, 22, 22, 64]\n",
        "    # Output Tensor Shape: [batch_size, 22 * 22 * 64]\n",
        "    pool2_flat = tf.reshape(bn2, [-1, 22 * 22 * 64])\n",
        "\n",
        "    # FC Layer with 1024 neurons\n",
        "    fc1 = tf.contrib.layers.fully_connected(inputs = pool2_flat, num_outputs = 1024, activation_fn = None, \n",
        "                                           weights_initializer=tf.contrib.layers.xavier_initializer())\n",
        "    \n",
        "    # BatchNorm Layer\n",
        "    bn3 = tf.contrib.layers.batch_norm(inputs = fc1, activation_fn = tf.nn.relu)\n",
        "\t\n",
        "    # Dropout Layer (0.8 probability of keeping the element)\n",
        "    dropout1 = tf.contrib.layers.dropout(inputs = bn3, keep_prob = 0.8, is_training = (mode == tf.estimator.ModeKeys.TRAIN))\n",
        "\t\n",
        "    # FC Layer with 1024 neurons\n",
        "    fc2 = tf.contrib.layers.fully_connected(inputs = dropout1, num_outputs = 1024, activation_fn = None, \n",
        "                                           weights_initializer=tf.contrib.layers.xavier_initializer())\n",
        "    \n",
        "    # BatchNorm Layer\n",
        "    bn4 = tf.contrib.layers.batch_norm(inputs = fc2, activation_fn = tf.nn.relu)\n",
        "    \n",
        "    # Dropout Layer (0.8 probability of keeping the element)\n",
        "    dropout2 = tf.contrib.layers.dropout(inputs = bn4, keep_prob = 0.8, is_training = (mode == tf.estimator.ModeKeys.TRAIN))\n",
        "    \n",
        "    # FC Layer with 1024 neurons\n",
        "    fc3 = tf.contrib.layers.fully_connected(inputs = dropout2, num_outputs = 1024, activation_fn = None, \n",
        "                                           weights_initializer=tf.contrib.layers.xavier_initializer())\n",
        "    \n",
        "    # BatchNorm Layer\n",
        "    bn5 = tf.contrib.layers.batch_norm(inputs = fc3, activation_fn = tf.nn.relu)\n",
        "\n",
        "    # Dropout Layer (0.8 probability of keeping the element)\n",
        "    dropout3 = tf.contrib.layers.dropout(inputs = bn5, keep_prob = 0.8, is_training = (mode == tf.estimator.ModeKeys.TRAIN))\n",
        "    \n",
        "    # Regression Layer\n",
        "    # Input Tensor Shape: [batch_size, 1024]\n",
        "    # Output Tensor Shape: [batch_size, 1]\n",
        "    regression = tf.contrib.layers.fully_connected(inputs = dropout3, num_outputs = 1, activation_fn = None)\n",
        "    \n",
        "    if mode == tf.estimator.ModeKeys.PREDICT:\n",
        "      # Round the predictions\n",
        "      return tf.estimator.EstimatorSpec(mode = mode, predictions = tf.math.round(regression))\n",
        "    \n",
        "    # Loss function\n",
        "    # loss = tf.losses.mean_squared_error(labels = labels, predictions = regression)\n",
        "    loss = tf.losses.absolute_difference(labels = labels, predictions = regression)\n",
        "    loss_to_log = tf.identity(loss, name=\"loss_to_log\")\n",
        "    \n",
        "    # For TensorBoard Visualization\n",
        "    tf.summary.scalar('loss', loss)\n",
        "    \n",
        "    # Adam Optimizer\n",
        "    if mode == tf.estimator.ModeKeys.TRAIN:\n",
        "      optimizer = tf.train.AdamOptimizer(learning_rate = 0.001, beta1 = 0.9, beta2 = 0.999, epsilon = 1e-08, use_locking = False, name = 'Adam')\n",
        "      train_op = optimizer.minimize(\n",
        "          loss = loss,\n",
        "          global_step = tf.train.get_global_step())\n",
        "      return tf.estimator.EstimatorSpec(mode = mode, loss = loss, train_op = train_op)\n",
        "\n",
        "    eval_metric_ops = {\n",
        "        \"MAE\": tf.metrics.mean_absolute_error(labels = labels, predictions = regression)\n",
        "    }\n",
        "    MAE_to_log = tf.identity(eval_metric_ops[\"MAE\"], name=\"MAE_to_log\")\n",
        "    \n",
        "    # For TensorBoard Visualization\n",
        "    tf.summary.scalar('MAE', MAE_to_log)\n",
        "    \n",
        "    return tf.estimator.EstimatorSpec(mode = mode, loss = loss, eval_metric_ops = eval_metric_ops)\n",
        "\n",
        "def main():\n",
        "    # Load the data\n",
        "    # Use below code to re-generate numpy files from images\n",
        "    # training_data, training_labels, validation_data, validation_labels, test_data, test_labels = _load_dataset()\n",
        "    \n",
        "    # For colab\n",
        "    #'''\n",
        "    training_data = np.load(\"/content/drive/My Drive/Colab Notebooks/training_data.npy\")\n",
        "    training_labels = np.load(\"/content/drive/My Drive/Colab Notebooks/training_labels.npy\")\n",
        "    validation_data = np.load(\"/content/drive/My Drive/Colab Notebooks/validation_data.npy\")\n",
        "    validation_labels = np.load(\"/content/drive/My Drive/Colab Notebooks/validation_labels.npy\")\n",
        "    test_data = np.load(\"/content/drive/My Drive/Colab Notebooks/test_data.npy\")\n",
        "    test_labels = np.load(\"/content/drive/My Drive/Colab Notebooks/test_labels.npy\")\n",
        "    #'''\n",
        "    \n",
        "    # For local \n",
        "    '''\n",
        "    training_data = np.load(\"training_data.npy\")\n",
        "    training_labels = np.load(\"training_labels.npy\")\n",
        "    validation_data = np.load(\"validation_data.npy\")\n",
        "    validation_labels = np.load(\"validation_labels.npy\")\n",
        "    test_data = np.load(\"test_data.npy\")\n",
        "    test_labels = np.load(\"test_labels.npy\")\n",
        "    '''\n",
        "    \n",
        "    config = tf.estimator.RunConfig( model_dir='/content/drive/My Drive/Colab Notebooks/savedmodels/final_model_dir',\n",
        "                                     keep_checkpoint_max=2,\n",
        "                                     save_checkpoints_steps=1000,\n",
        "                                     save_checkpoints_secs=None)\n",
        "    \n",
        "    # Estimator\n",
        "    age_estimator = tf.estimator.Estimator(model_fn = cnn_model, config = config)\n",
        "    \n",
        "    train_input_fn = tf.estimator.inputs.numpy_input_fn(\n",
        "        x = {\"x\": training_data},\n",
        "        y = training_labels,\n",
        "        batch_size = 32,\n",
        "        num_epochs = None,\n",
        "        shuffle = True\n",
        "    )\n",
        "    \n",
        "    eval_logging_hook = tf.train.LoggingTensorHook(tensors={\"MAE_to_log\": \"MAE_to_log\"}, every_n_iter = 1)\n",
        "    \n",
        "    '''\n",
        "    # Training and evaluation (early stopping based on validation MAE)\n",
        "    for i in range(100000):\n",
        "      age_estimator.train(\n",
        "          input_fn = train_input_fn,\n",
        "          steps = 100\n",
        "          #,hooks = [logging_hook]\n",
        "      )\n",
        "\n",
        "      # Evaluate the model and print results\n",
        "      eval_input_fn = tf.estimator.inputs.numpy_input_fn(x={\"x\": validation_data}, y = validation_labels, num_epochs = 1, shuffle = False)\n",
        "\n",
        "      eval_results = age_estimator.evaluate(input_fn = eval_input_fn, hooks = [eval_logging_hook], steps = 1)\n",
        "      print(eval_results)\n",
        "    \n",
        "      mae = eval_results[\"MAE\"]\n",
        "      \n",
        "      # Early stopping based on validation MAE\n",
        "      if mae < 5.75:\n",
        "        break\n",
        "    '''    \n",
        "    \n",
        "    # Test (comment out the block above when testing)\n",
        "    test_input_fn = tf.estimator.inputs.numpy_input_fn(x={\"x\": test_data}, y = test_labels, num_epochs = 1, shuffle = False)\n",
        "    \n",
        "    # For report only\n",
        "    #test_input_fn = tf.estimator.inputs.numpy_input_fn(x={\"x\": validation_data}, y = validation_labels, num_epochs = 1, shuffle = False)\n",
        "    #test_input_fn = tf.estimator.inputs.numpy_input_fn(x={\"x\": training_data}, y = training_labels, num_epochs = 1, shuffle = False)\n",
        "    \n",
        "    test_results = age_estimator.evaluate(input_fn = test_input_fn, hooks = [eval_logging_hook], steps = 1)\n",
        "    print(test_results)\n",
        "    \n",
        "    # Informative success/failure examples\n",
        "    sample_predictions = age_estimator.predict(input_fn = test_input_fn)\n",
        "    pred_list = np.array(list(sample_predictions))\n",
        "    pred_list2 = pred_list\n",
        "    failure_index = np.argmax(np.absolute(np.subtract(pred_list, test_labels)))\n",
        "    success_index = np.argmin(np.absolute(np.subtract(pred_list2, test_labels)))\n",
        "    \n",
        "    print(\"Failure example\")\n",
        "    failure = plt.imshow(test_data[failure_index], cmap=\"gray\")\n",
        "    plt.show()\n",
        "    print(\"Predicted: \", int(pred_list[failure_index]), \", Real: \", int(test_labels[failure_index]))\n",
        "    \n",
        "    print(\"\\nSuccess example\")\n",
        "    success = plt.imshow(test_data[success_index], cmap=\"gray\")\n",
        "    plt.show()\n",
        "    print(\"Predicted: \", int(pred_list[success_index]), \", Real: \", int(test_labels[success_index]))\n",
        "    \n",
        "def _load_dataset():\n",
        "    # Training set\n",
        "    training_data = []\n",
        "    training_labels = []\n",
        "    \n",
        "    tr_path = 'UTKFace_downsampled/training_set'\n",
        "    for filename in os.listdir(tr_path):\n",
        "        training_labels.append(int(filename[:3]))\n",
        "        img_data = cv2.imread(os.path.join(tr_path, filename), cv2.IMREAD_GRAYSCALE).astype(np.float32)\n",
        "        img_data = cv2.resize(img_data, (88, 88)) # Resize\n",
        "        training_data.append(img_data)\n",
        "        \n",
        "    training_data = np.array(training_data, dtype='float32')\n",
        "    training_labels = np.array(training_labels, dtype='float32')\n",
        "    training_labels = training_labels.reshape(len(training_labels), 1)\n",
        "    np.save('training_data', training_data)\n",
        "    np.save('training_labels', training_labels)\n",
        "    \n",
        "    # Validation set\n",
        "    validation_data = []\n",
        "    validation_labels = []\n",
        "    \n",
        "    v_path = 'UTKFace_downsampled/validation_set'\n",
        "    for filename in os.listdir(v_path):\n",
        "        validation_labels.append(int(filename[:3]))\n",
        "        img_data = cv2.imread(os.path.join(v_path, filename), cv2.IMREAD_GRAYSCALE).astype(np.float32)\n",
        "        img_data = cv2.resize(img_data, (88, 88)) # Resize\n",
        "        validation_data.append(img_data)\n",
        "        \n",
        "    validation_data = np.array(validation_data, dtype='float32')\n",
        "    validation_labels = np.array(validation_labels, dtype='float32')\n",
        "    validation_labels = validation_labels.reshape(len(validation_labels), 1)\n",
        "    np.save('validation_data', validation_data)\n",
        "    np.save('validation_labels', validation_labels)\n",
        "    \n",
        "    # Test set\n",
        "    test_data = []\n",
        "    test_labels = []\n",
        "    \n",
        "    t_path = 'UTKFace_downsampled/test_set'\n",
        "    for filename in os.listdir(t_path):\n",
        "        test_labels.append(int(filename[:3]))\n",
        "        img_data = cv2.imread(os.path.join(t_path, filename), cv2.IMREAD_GRAYSCALE).astype(np.float32)\n",
        "        img_data = cv2.resize(img_data, (88, 88)) # Resize\n",
        "        test_data.append(img_data)\n",
        "        \n",
        "    test_data = np.array(test_data, dtype='float32')\n",
        "    test_labels = np.array(test_labels, dtype='float32')\n",
        "    test_labels = test_labels.reshape(len(test_labels), 1)\n",
        "    np.save('test_data', test_data)\n",
        "    np.save('test_labels', test_labels)\n",
        "    \n",
        "    return training_data, training_labels, validation_data, validation_labels, test_data, test_labels\n",
        "    \n",
        "if __name__ == \"__main__\":\n",
        "  main()"
      ],
      "execution_count": 0,
      "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",
            "INFO:tensorflow:Using config: {'_model_dir': '/content/drive/My Drive/Colab Notebooks/savedmodels/final_model_dir', '_tf_random_seed': None, '_save_summary_steps': 100, '_save_checkpoints_steps': 1000, '_save_checkpoints_secs': None, '_session_config': allow_soft_placement: true\n",
            "graph_options {\n",
            "  rewrite_options {\n",
            "    meta_optimizer_iterations: ONE\n",
            "  }\n",
            "}\n",
            ", '_keep_checkpoint_max': 2, '_keep_checkpoint_every_n_hours': 10000, '_log_step_count_steps': 100, '_train_distribute': None, '_device_fn': None, '_protocol': None, '_eval_distribute': None, '_experimental_distribute': None, '_service': None, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7fd5cd9aff28>, '_task_type': 'worker', '_task_id': 0, '_global_id_in_cluster': 0, '_master': '', '_evaluation_master': '', '_is_chief': True, '_num_ps_replicas': 0, '_num_worker_replicas': 1}\n",
            "INFO:tensorflow:Calling model_fn.\n",
            "INFO:tensorflow:Done calling model_fn.\n",
            "INFO:tensorflow:Starting evaluation at 2019-04-07T10:20:03Z\n",
            "INFO:tensorflow:Graph was finalized.\n",
            "INFO:tensorflow:Restoring parameters from /content/drive/My Drive/Colab Notebooks/savedmodels/final_model_dir/model.ckpt-26800\n",
            "INFO:tensorflow:Running local_init_op.\n",
            "INFO:tensorflow:Done running local_init_op.\n",
            "INFO:tensorflow:MAE_to_log = [0.        5.6883373]\n",
            "INFO:tensorflow:Evaluation [1/1]\n",
            "INFO:tensorflow:Finished evaluation at 2019-04-07-10:20:04\n",
            "INFO:tensorflow:Saving dict for global step 26800: MAE = 5.6883373, global_step = 26800, loss = 5.6883373\n",
            "INFO:tensorflow:Saving 'checkpoint_path' summary for global step 26800: /content/drive/My Drive/Colab Notebooks/savedmodels/final_model_dir/model.ckpt-26800\n",
            "{'MAE': 5.6883373, 'loss': 5.6883373, 'global_step': 26800}\n",
            "INFO:tensorflow:Calling model_fn.\n",
            "INFO:tensorflow:Done calling model_fn.\n",
            "INFO:tensorflow:Graph was finalized.\n",
            "INFO:tensorflow:Restoring parameters from /content/drive/My Drive/Colab Notebooks/savedmodels/final_model_dir/model.ckpt-26800\n",
            "INFO:tensorflow:Running local_init_op.\n",
            "INFO:tensorflow:Done running local_init_op.\n",
            "Failure example\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUsAAAFMCAYAAABCsp4mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/QAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Predicted:  32 , Real:  76\n",
            "\n",
            "Success example\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUsAAAFMCAYAAABCsp4mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Predicted:  1 , Real:  1\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "8JgY9_6Td4K-",
        "colab_type": "code",
        "colab": {}
      },
      "cell_type": "code",
      "source": [
        "# For downloading the model as a zip file\n",
        "import shutil as sh\n",
        "sh.make_archive(\"/content/drive/My Drive/Colab Notebooks/final_model_dir\", 'zip', '/content/drive/My Drive/Colab Notebooks/savedmodels/final_model_dir')"
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}