{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "dedc2602",
   "metadata": {},
   "source": [
    "# Creating a convolutional network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "701fb5bd",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv2d (Conv2D)             (None, 228, 150, 20)      1520      \n",
      "                                                                 \n",
      " dropout (Dropout)           (None, 228, 150, 20)      0         \n",
      "                                                                 \n",
      " conv2d_1 (Conv2D)           (None, 224, 146, 20)      10020     \n",
      "                                                                 \n",
      " dropout_1 (Dropout)         (None, 224, 146, 20)      0         \n",
      "                                                                 \n",
      " max_pooling2d (MaxPooling2D  (None, 74, 48, 20)       0         \n",
      " )                                                               \n",
      "                                                                 \n",
      " conv2d_2 (Conv2D)           (None, 70, 44, 20)        10020     \n",
      "                                                                 \n",
      " dropout_2 (Dropout)         (None, 70, 44, 20)        0         \n",
      "                                                                 \n",
      " conv2d_3 (Conv2D)           (None, 66, 40, 10)        5010      \n",
      "                                                                 \n",
      " dropout_3 (Dropout)         (None, 66, 40, 10)        0         \n",
      "                                                                 \n",
      " max_pooling2d_1 (MaxPooling  (None, 22, 13, 10)       0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " flatten (Flatten)           (None, 2860)              0         \n",
      "                                                                 \n",
      " dense (Dense)               (None, 4)                 11444     \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 38,014\n",
      "Trainable params: 38,014\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow.keras import models, layers\n",
    "\n",
    "conv_network = models.Sequential()\n",
    "conv_network.add(layers.Conv2D(20, (5,5), activation='relu', input_shape=(232, 154, 3)))\n",
    "conv_network.add(layers.Dropout(0.2))\n",
    "conv_network.add(layers.Conv2D(20, (5,5), activation='relu'))\n",
    "conv_network.add(layers.Dropout(0.2))\n",
    "conv_network.add(layers.MaxPooling2D(3,3))\n",
    "conv_network.add(layers.Conv2D(20, (5,5), activation='relu'))\n",
    "conv_network.add(layers.Dropout(0.2))\n",
    "conv_network.add(layers.Conv2D(10, (5,5), activation='relu'))\n",
    "conv_network.add(layers.Dropout(0.2))\n",
    "conv_network.add(layers.MaxPooling2D(3,3))\n",
    "conv_network.add(layers.Flatten())\n",
    "conv_network.add(layers.Dense(4, activation='softmax'))\n",
    "\n",
    "optimizer=tf.keras.optimizers.Adam(learning_rate=0.02)\n",
    "\n",
    "conv_network.compile(optimizer=optimizer, loss='mse', metrics=['accuracy'])\n",
    "\n",
    "conv_network.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "4ab96d93",
   "metadata": {},
   "source": [
    "# Loading in the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "2a6353d7",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 2008 files belonging to 4 classes.\n",
      "Using 1607 files for training.\n",
      "Found 2008 files belonging to 4 classes.\n",
      "Using 401 files for validation.\n"
     ]
    }
   ],
   "source": [
    "data_dir = \"/Users/kerickwalker/src/dis/deep_learning/bat_data\"\n",
    "\n",
    "img_width = 154\n",
    "img_height = 232\n",
    "batch_size = 128\n",
    "\n",
    "# Load in the training data\n",
    "training_data = tf.keras.utils.image_dataset_from_directory(\n",
    "  data_dir,\n",
    "  validation_split=0.2,\n",
    "  subset=\"training\",\n",
    "  seed=123,\n",
    "  image_size=(img_height, img_width),\n",
    "  batch_size=batch_size)\n",
    "\n",
    "# Load in validation data\n",
    "validation_data = tf.keras.utils.image_dataset_from_directory(\n",
    "  data_dir,\n",
    "  validation_split=0.2,\n",
    "  subset=\"validation\",\n",
    "  seed=123,\n",
    "  image_size=(img_height, img_width),\n",
    "  batch_size=batch_size)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cd4adeaa",
   "metadata": {},
   "source": [
    "# Training convolutional network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c1d53cef",
   "metadata": {},
   "outputs": [],
   "source": [
    "conv_network.fit(training_data, validation_data=validation_data, epochs=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8a22d520",
   "metadata": {},
   "source": [
    "# Transfer Learning with MobileNetV2"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7451e896",
   "metadata": {},
   "source": [
    "#### Convert dataset to numpy array for preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "32c2dd65",
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "from tensorflow.keras.applications import MobileNetV2\n",
    "from tensorflow.keras import layers, models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "bcff2372",
   "metadata": {},
   "outputs": [],
   "source": [
    "img_size = (232, 154)  # MobileNetV2 input size\n",
    "batch_size = 32\n",
    "data_dir = \"/Users/kerickwalker/src/dis/deep_learning/bat_data\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "26d31c9f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 2008 images belonging to 4 classes.\n"
     ]
    }
   ],
   "source": [
    "train_datagen = ImageDataGenerator(\n",
    "    rescale=1./255,\n",
    "    rotation_range=20,\n",
    "    width_shift_range=0.2,\n",
    "    height_shift_range=0.2,\n",
    "    shear_range=0.2,\n",
    "    zoom_range=0.2,\n",
    "    horizontal_flip=True,\n",
    "    fill_mode='nearest'\n",
    ")\n",
    "\n",
    "train_generator = train_datagen.flow_from_directory(\n",
    "    data_dir,\n",
    "    target_size=img_size,\n",
    "    batch_size=batch_size,\n",
    "    class_mode='categorical',\n",
    "    shuffle=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "cf420374",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:`input_shape` is undefined or non-square, or `rows` is not in [96, 128, 160, 192, 224]. Weights for input shape (224, 224) will be loaded as the default.\n"
     ]
    }
   ],
   "source": [
    "base_model = MobileNetV2(\n",
    "    input_shape=(232, 154, 3),\n",
    "    include_top=False,\n",
    "    weights='imagenet'\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "e7e027fb",
   "metadata": {},
   "outputs": [],
   "source": [
    "for layer in base_model.layers:\n",
    "    layer.trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "2bd9014d",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = models.Sequential()\n",
    "model.add(base_model)\n",
    "model.add(layers.GlobalAveragePooling2D())\n",
    "model.add(layers.Dense(256, activation='relu'))\n",
    "model.add(layers.Dropout(0.5))\n",
    "model.add(layers.Dense(4, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "04aef745",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(\n",
    "    optimizer='adam',\n",
    "    loss='categorical_crossentropy',\n",
    "    metrics=['accuracy']\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "4f624f89",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-11-30 18:29:04.053048: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_0' with dtype int32\n",
      "\t [[{{node Placeholder/_0}}]]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "62/62 [==============================] - 38s 560ms/step - loss: 0.6074 - accuracy: 0.7657\n",
      "Epoch 2/10\n",
      "62/62 [==============================] - 44s 715ms/step - loss: 0.2596 - accuracy: 0.9018\n",
      "Epoch 3/10\n",
      "62/62 [==============================] - 50s 809ms/step - loss: 0.2202 - accuracy: 0.9165\n",
      "Epoch 4/10\n",
      "62/62 [==============================] - 52s 833ms/step - loss: 0.1985 - accuracy: 0.9276\n",
      "Epoch 5/10\n",
      "62/62 [==============================] - 51s 822ms/step - loss: 0.1963 - accuracy: 0.9276\n",
      "Epoch 6/10\n",
      "62/62 [==============================] - 57s 922ms/step - loss: 0.2040 - accuracy: 0.9236\n",
      "Epoch 7/10\n",
      "62/62 [==============================] - 57s 912ms/step - loss: 0.1698 - accuracy: 0.9357\n",
      "Epoch 8/10\n",
      "62/62 [==============================] - 52s 834ms/step - loss: 0.1672 - accuracy: 0.9332\n",
      "Epoch 9/10\n",
      "62/62 [==============================] - 50s 795ms/step - loss: 0.1603 - accuracy: 0.9408\n",
      "Epoch 10/10\n",
      "62/62 [==============================] - 48s 778ms/step - loss: 0.1711 - accuracy: 0.9332\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(\n",
    "    train_generator,\n",
    "    steps_per_epoch=train_generator.samples // batch_size,\n",
    "    epochs=10\n",
    ")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "disdl",
   "language": "python",
   "name": "disdl"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}