model_builder.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import os\n",
    "import PIL\n",
    "import PIL.Image\n",
    "import math\n",
    "import random\n",
    "import tensorflow as tf\n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib import image as mpimg\n",
    "from  tensorflow.keras.preprocessing.image import ImageDataGenerator"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Import packages and check versions "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensorflow version: 2.6.2\n"
     ]
    }
   ],
   "source": [
    "print('Tensorflow version:',tf.__version__)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Define Program Constants"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "DATA_DIR = './data/'\n",
    "IMAGE_HEIGHT = 224\n",
    "IMAGE_WIDTH = 224\n",
    "IMAGE_SHAPE = (IMAGE_HEIGHT,IMAGE_WIDTH)\n",
    "INPUT_LAYER_SHAPE = (IMAGE_HEIGHT,IMAGE_WIDTH, 3) #RGB color for last value\n",
    "IMAGE_PROCESSING_BATCH_SIZE = 32\n",
    "VALIDATION_SPLIT=0.2\n",
    "SEED=123\n",
    "BATCH_SIZE=128\n",
    "EPOCHS=5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Constrain the randomness\n",
    "random.seed(SEED)\n",
    "tf.random.set_seed(SEED)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3. Make image generation and formatting objects"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 60 images belonging to 2 classes.\n",
      "Found 14 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "img_data_gen = ImageDataGenerator(\n",
    "        rescale=1./255,\n",
    "        shear_range=0.2,\n",
    "        zoom_range=0.2,\n",
    "        horizontal_flip=True,\n",
    "        validation_split=VALIDATION_SPLIT)\n",
    "\n",
    "train_generator = img_data_gen.flow_from_directory(\n",
    "    DATA_DIR,\n",
    "    target_size=IMAGE_SHAPE,\n",
    "    batch_size=IMAGE_PROCESSING_BATCH_SIZE,\n",
    "    class_mode='categorical',\n",
    "    seed=SEED,\n",
    "    subset='training') # set as training data\n",
    "\n",
    "validation_generator = img_data_gen.flow_from_directory(\n",
    "    DATA_DIR, # same directory as training data\n",
    "    target_size=IMAGE_SHAPE,\n",
    "    batch_size=IMAGE_PROCESSING_BATCH_SIZE,\n",
    "    class_mode='categorical',\n",
    "    seed=SEED,\n",
    "    subset='validation') # set as validation data"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Sample Image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "images_paths = [os.path.join(path, name) for path, subdirs, files in os.walk(DATA_DIR) for name in files]\n",
    "\n",
    "image_path = images_paths[random.randint(0, len(images_paths) - 1)]\n",
    "\n",
    "image = mpimg.imread(image_path)     \n",
    "image_type=image_path.split('/')[3]\n",
    "plt.title('Image class: %s' % image_type)\n",
    "plt.imshow(image)\n",
    "plt.show()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4. Define hyperparameter grid search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorboard.plugins.hparams import api as hp\n",
    "\n",
    "HP_NUM_UNITS = hp.HParam('num_units', hp.Discrete([16]))\n",
    "HP_DROPOUT = hp.HParam('dropout', hp.Discrete([0.25]))\n",
    "HP_OPTIMIZER = hp.HParam('optimizer', hp.Discrete(['adam']))\n",
    "HP_LEARNING_RATE = hp.HParam('lr', hp.Discrete([1e-5]))\n",
    "HP_STRIDES = hp.HParam('strides', hp.Discrete([3]))\n",
    "HP_KERNEL_SIZE = hp.HParam('kernel_size', hp.Discrete([ 4, 5])) \n",
    "HP_POOL_SIZE = hp.HParam('pool_size', hp.Discrete([2]))"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5. Define Network Architecture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras import layers\n",
    "\n",
    "## Output layer interface\n",
    "OUTPUT_LAYER = layers.Dense(train_generator.num_classes, activation=\"softmax\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "from tensorflow import keras\n",
    "\n",
    "def build_model_architecture(hyper_params) -> keras.Model: \n",
    "\n",
    "    return keras.Sequential([\n",
    "                    keras.Input(shape=INPUT_LAYER_SHAPE),\n",
    "                    layers.Conv2D(32, kernel_size=hyper_params[HP_KERNEL_SIZE], activation=\"relu\"),\n",
    "                    layers.MaxPooling2D(pool_size=hyper_params[HP_POOL_SIZE], strides=hyper_params[HP_STRIDES]),\n",
    "                    layers.Conv2D(64, kernel_size=hyper_params[HP_KERNEL_SIZE], activation=\"relu\"),\n",
    "                    layers.MaxPooling2D(pool_size=hyper_params[HP_POOL_SIZE], strides=hyper_params[HP_STRIDES]),\n",
    "                    layers.Flatten(),\n",
    "                    layers.Dense(hyper_params[HP_NUM_UNITS], activation=tf.nn.relu),\n",
    "                    layers.Dropout(0.5),\n",
    "                    OUTPUT_LAYER,\n",
    "                ])\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6. Clean up log folder and start up Tensorboard"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "id_20012023181115\n",
      "Error when trying to remove log directory\n"
     ]
    }
   ],
   "source": [
    "try:\n",
    "    for dir in os.listdir('./logs/'):\n",
    "        print(dir)\n",
    "        os.removedirs(f'./logs/{dir}')\n",
    "except:\n",
    "    print('Error when trying to remove log directory')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The tensorboard extension is already loaded. To reload it, use:\n",
      "  %reload_ext tensorboard\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Reusing TensorBoard on port 6006 (pid 4644), started 2:27:05 ago. (Use '!kill 4644' to kill it.)"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/html": [
       "\n",
       "      <iframe id=\"tensorboard-frame-7f9950b069f7ba86\" width=\"100%\" height=\"800\" frameborder=\"0\">\n",
       "      </iframe>\n",
       "      <script>\n",
       "        (function() {\n",
       "          const frame = document.getElementById(\"tensorboard-frame-7f9950b069f7ba86\");\n",
       "          const url = new URL(\"http://localhost\");\n",
       "          const port = 6006;\n",
       "          if (port) {\n",
       "            url.port = port;\n",
       "          }\n",
       "          frame.src = url;\n",
       "        })();\n",
       "      </script>\n",
       "    "
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%load_ext tensorboard\n",
    "\n",
    "%tensorboard --logdir logs/"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 7. Run the training session"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--- Starting trial: id_20012023184328\n",
      "{'num_units': 16, 'dropout': 0.25, 'optimizer': 'adam', 'lr': 1e-05, 'strides': 3, 'kernel_size': 4, 'pool_size': 2}\n",
      "Epoch 1/5\n",
      "2/2 - 3s - loss: 0.7984 - accuracy: 0.2667 - val_loss: 0.7543 - val_accuracy: 0.2857\n",
      "Epoch 2/5\n",
      "2/2 - 2s - loss: 0.7557 - accuracy: 0.3000 - val_loss: 0.7071 - val_accuracy: 0.3571\n",
      "Epoch 3/5\n",
      "2/2 - 2s - loss: 0.6930 - accuracy: 0.5833 - val_loss: 0.6741 - val_accuracy: 0.7143\n",
      "Epoch 4/5\n",
      "2/2 - 2s - loss: 0.6765 - accuracy: 0.5833 - val_loss: 0.6465 - val_accuracy: 0.7143\n",
      "Epoch 5/5\n",
      "2/2 - 2s - loss: 0.6340 - accuracy: 0.6500 - val_loss: 0.6246 - val_accuracy: 0.7143\n",
      "INFO:tensorflow:Assets written to: logs/id_20012023184328/model/assets\n",
      "--- Starting trial: id_20012023184340\n",
      "{'num_units': 16, 'dropout': 0.25, 'optimizer': 'adam', 'lr': 1e-05, 'strides': 3, 'kernel_size': 5, 'pool_size': 2}\n",
      "Epoch 1/5\n",
      "2/2 - 3s - loss: 0.7055 - accuracy: 0.4333 - val_loss: 0.7022 - val_accuracy: 0.3571\n",
      "Epoch 2/5\n",
      "2/2 - 2s - loss: 0.6951 - accuracy: 0.5000 - val_loss: 0.6855 - val_accuracy: 0.6429\n",
      "Epoch 3/5\n",
      "2/2 - 2s - loss: 0.6686 - accuracy: 0.6500 - val_loss: 0.6702 - val_accuracy: 0.7143\n",
      "Epoch 4/5\n",
      "2/2 - 2s - loss: 0.6530 - accuracy: 0.7333 - val_loss: 0.6555 - val_accuracy: 0.7143\n",
      "Epoch 5/5\n",
      "2/2 - 2s - loss: 0.6533 - accuracy: 0.6667 - val_loss: 0.6427 - val_accuracy: 0.7143\n",
      "INFO:tensorflow:Assets written to: logs/id_20012023184340/model/assets\n"
     ]
    }
   ],
   "source": [
    "from tensorflow.keras.callbacks import EarlyStopping\n",
    "from datetime import datetime\n",
    "\n",
    "assert HP_NUM_UNITS.domain is not None, 'HP_NUM_UNITS has no \"Domain\" attribute'\n",
    "for num_units in HP_NUM_UNITS.domain.values:\n",
    "\n",
    "    assert HP_DROPOUT.domain is not None, 'HP_DROPOUT has no \"Domain\" attribute'\n",
    "    # for dropout_rate in (HP_DROPOUT.domain.min_value, HP_DROPOUT.domain.max_value):\n",
    "    for dropout_rate in HP_DROPOUT.domain.values:\n",
    "    \n",
    "        assert HP_OPTIMIZER.domain is not None, 'HP_OPTIMIZER has no \"Domain\" attribute'\n",
    "        for optimizer in HP_OPTIMIZER.domain.values:\n",
    "        \n",
    "            assert HP_LEARNING_RATE.domain is not None, 'HP_LEARNING_RATE has no \"Domain\" attribute'\n",
    "            for learning_rate in HP_LEARNING_RATE.domain.values:\n",
    "                \n",
    "                assert HP_STRIDES.domain is not None, 'HP_STRIDES has no \"Domain\" attribute'\n",
    "                for strides in HP_STRIDES.domain.values:\n",
    "               \n",
    "                    assert HP_KERNEL_SIZE.domain is not None, 'HP_KERNEL_SIZE has no \"Domain\" attribute'\n",
    "                    for kernel_size in HP_KERNEL_SIZE.domain.values:\n",
    "                    \n",
    "                        assert HP_POOL_SIZE.domain is not None, 'HP_POOL_SIZE has no \"Domain\" attribute'\n",
    "                        for pool_size in HP_POOL_SIZE.domain.values:\n",
    "                            \n",
    "                            run_id = f'id_{datetime.now().strftime(\"%d%m%Y%H%M%S\")}'.replace(' ', '_')\n",
    "                            \n",
    "                            assert type(optimizer) == str\n",
    "\n",
    "                            hparams = {\n",
    "                                HP_NUM_UNITS: num_units,\n",
    "                                HP_DROPOUT: dropout_rate,\n",
    "                                HP_OPTIMIZER: optimizer,\n",
    "                                HP_LEARNING_RATE: learning_rate,\n",
    "                                HP_STRIDES: strides,\n",
    "                                HP_KERNEL_SIZE: kernel_size,\n",
    "                                HP_POOL_SIZE: pool_size\n",
    "                            }\n",
    "                            print('--- Starting trial: %s' % run_id)\n",
    "\n",
    "                            print({h.name: hparams[h] for h in hparams})\n",
    "\n",
    "                            run_dir = f'logs/{run_id}'\n",
    "                            \n",
    "                            op = keras.optimizers.Adam(learning_rate=learning_rate) if optimizer == 'adam' else keras.optimizers.SGD(learning_rate=learning_rate)\n",
    "                            \n",
    "                            model = build_model_architecture(hparams)\n",
    "\n",
    "                            model.compile(\n",
    "                                loss=\"categorical_crossentropy\",    \n",
    "                                optimizer=op, metrics=[\"accuracy\"]\n",
    "                            )\n",
    "\n",
    "                            # fit the model\n",
    "                            history = model.fit(\n",
    "                                train_generator,\n",
    "                                validation_data=validation_generator,\n",
    "                                batch_size=BATCH_SIZE,\n",
    "                                epochs=EPOCHS,\n",
    "                                verbose=2,\n",
    "                                callbacks=[\n",
    "                                    EarlyStopping(\n",
    "                                        monitor='loss',\n",
    "                                        mode='min',\n",
    "                                        patience=5,\n",
    "                                        min_delta=0.001,\n",
    "                                        restore_best_weights=True\n",
    "                                    ),\n",
    "                                    tf.keras.callbacks.TensorBoard(\n",
    "                                        run_dir, histogram_freq=1),  # log metrics\n",
    "                                    hp.KerasCallback(run_dir, hparams),  # log hparams\n",
    "                                ]\n",
    "                            )\n",
    "\n",
    "                            # Evaluate on Validation data\n",
    "                            # loss, acc = model.evaluate(validation_generator)\n",
    "                            # print(\"Val loss:\", loss, \"Accuracy\",acc)\n",
    "                            \n",
    "                            model.save(run_dir + '/model')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.6.13 ('spiced_deep_learning')",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.13"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "e3c2b689d07e7d9b2d2315b383ad3b57f8181e1eee1eccb32fbd62e20f14292f"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}