training.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import tensorflow core\n",
    "\n",
    "import tensorflow as tf\n",
    "\n",
    "# Import Keras Model and Layer\n",
    "\n",
    "from tensorflow.keras.applications import MobileNetV2\n",
    "from tensorflow.keras.models import Model\n",
    "from tensorflow.keras.layers import AveragePooling2D\n",
    "from tensorflow.keras.layers import Dropout\n",
    "from tensorflow.keras.layers import Flatten\n",
    "from tensorflow.keras.layers import Dense\n",
    "from tensorflow.keras.layers import Input\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "\n",
    "# Import Keras Utility\n",
    "\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "from tensorflow.keras.preprocessing.image import img_to_array\n",
    "from tensorflow.keras.preprocessing.image import load_img\n",
    "from tensorflow.keras.utils import to_categorical\n",
    "from tensorflow.keras.callbacks import EarlyStopping\n",
    "\n",
    "# Import library for utility\n",
    "\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.model_selection import train_test_split\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.gridspec as gridspec\n",
    "import glob\n",
    "import numpy as np\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 864x288 with 12 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualize Dataset\n",
    "\n",
    "dataset_paths = glob.glob(\"./data/source/**/*.jpg\")\n",
    "\n",
    "ncols = 6\n",
    "nrows = 2\n",
    "figure = plt.figure(figsize=(12, 4))\n",
    "plt.rcParams.update({\"font.size\": 14})\n",
    "grid = gridspec.GridSpec(ncols=ncols, nrows=nrows, figure=figure)\n",
    "\n",
    "np.random.seed(0)\n",
    "rand = np.random.randint(0, len(dataset_paths), size=(ncols * nrows))\n",
    "\n",
    "for i in range(ncols * nrows):\n",
    "    figure.add_subplot(grid[i])\n",
    "    image = load_img(dataset_paths[rand[i]])\n",
    "    label = dataset_paths[rand[i]].split(os.path.sep)[-2]\n",
    "    plt.title('\"{:s}\"'.format(label))\n",
    "    plt.axis(False)\n",
    "    plt.imshow(image)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[INFO] Found 37623 images with 36 classes\n"
     ]
    }
   ],
   "source": [
    "# Label Encoding\n",
    "\n",
    "X = []\n",
    "labels = []\n",
    "\n",
    "for image_path in dataset_paths:\n",
    "    label = image_path.split(os.path.sep)[-2]\n",
    "    image = load_img(image_path, target_size=(80, 80))\n",
    "    image = img_to_array(image)\n",
    "\n",
    "    X.append(image)\n",
    "    labels.append(label)\n",
    "\n",
    "X = np.array(X,dtype=\"float16\")\n",
    "labels = np.array(labels)\n",
    "\n",
    "print(\"[INFO] Found {:d} images with {:d} classes\".format(len(X), len(set(labels))))\n",
    "\n",
    "# One-hot encoding\n",
    "lb = LabelEncoder()\n",
    "lb.fit(labels)\n",
    "labels = lb.transform(labels)\n",
    "y = to_categorical(labels)\n",
    "\n",
    "np.save('license_character_classes.npy', lb.classes_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "(train_x, test_x, train_y, test_y) = train_test_split(X, y, test_size=0.1, stratify=y, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "def construct_model(output_shape):\n",
    "    base_model = MobileNetV2(\n",
    "        weights=\"imagenet\", include_top=False, input_tensor=Input(shape=(80, 80, 3)))\n",
    "\n",
    "    for layer in base_model.layers:\n",
    "        layer.trainable = True\n",
    "\n",
    "    base_model_output = base_model.output\n",
    "    fc_model = AveragePooling2D(pool_size=(3, 3))(base_model_output)\n",
    "    fc_model = Flatten(name=\"flatten\")(fc_model)\n",
    "    fc_model = Dense(128, activation=\"relu\")(fc_model)\n",
    "    fc_model = Dropout(0.5)(fc_model)\n",
    "    fc_model = Dense(output_shape, activation=\"softmax\")(fc_model)\n",
    "\n",
    "    return Model(inputs=base_model.input, outputs=fc_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"\n",
    "Model 01\n",
    "LR = 1e-3, EPOCH = 15, Testing data = 10%\n",
    "\n",
    "Model 02\n",
    "LR = 1e-4, EPOCH = 30, Testing data = 10%\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "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": [
    "# Construct Model\n",
    "LR = 1e-3\n",
    "EPOCHS = 15\n",
    "\n",
    "classes = y.shape[1]\n",
    "\n",
    "model = construct_model(output_shape=classes)\n",
    "\n",
    "optimizer = Adam(learning_rate=LR, decay=LR/EPOCHS)\n",
    "model.compile(loss=\"categorical_crossentropy\",\n",
    "              optimizer=optimizer, metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "augmentation = ImageDataGenerator(\n",
    "    rotation_range=15,\n",
    "    width_shift_range=0.1,\n",
    "    height_shift_range=0.1,\n",
    "    shear_range=0.1,\n",
    "    zoom_range=0.15,\n",
    "    horizontal_flip=True,\n",
    "    fill_mode=\"nearest\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/15\n",
      "529/529 [==============================] - 229s 421ms/step - loss: 0.9853 - accuracy: 0.7393 - val_loss: 8.7997 - val_accuracy: 0.0904\n",
      "Epoch 2/15\n",
      "529/529 [==============================] - 214s 403ms/step - loss: 0.4448 - accuracy: 0.8841 - val_loss: 3.4813 - val_accuracy: 0.4417\n",
      "Epoch 3/15\n",
      "529/529 [==============================] - 223s 421ms/step - loss: 0.3413 - accuracy: 0.9071 - val_loss: 0.9690 - val_accuracy: 0.8212\n",
      "Epoch 4/15\n",
      "529/529 [==============================] - 242s 457ms/step - loss: 0.3059 - accuracy: 0.9168 - val_loss: 1.0130 - val_accuracy: 0.7853\n",
      "Epoch 5/15\n",
      "529/529 [==============================] - 286s 540ms/step - loss: 0.2710 - accuracy: 0.9244 - val_loss: 0.5700 - val_accuracy: 0.8743\n",
      "Epoch 6/15\n",
      "529/529 [==============================] - 248s 468ms/step - loss: 0.2493 - accuracy: 0.9306 - val_loss: 0.5355 - val_accuracy: 0.8921\n",
      "Epoch 7/15\n",
      "529/529 [==============================] - 178s 337ms/step - loss: 0.2379 - accuracy: 0.9332 - val_loss: 0.5133 - val_accuracy: 0.9006\n",
      "Epoch 8/15\n",
      "529/529 [==============================] - 183s 347ms/step - loss: 0.2362 - accuracy: 0.9338 - val_loss: 0.5814 - val_accuracy: 0.8754\n",
      "Epoch 9/15\n",
      "529/529 [==============================] - 186s 351ms/step - loss: 0.2159 - accuracy: 0.9385 - val_loss: 1.2092 - val_accuracy: 0.8108\n",
      "Epoch 10/15\n",
      "529/529 [==============================] - 192s 363ms/step - loss: 0.2125 - accuracy: 0.9393 - val_loss: 0.2676 - val_accuracy: 0.9386\n",
      "Epoch 11/15\n",
      "529/529 [==============================] - 195s 368ms/step - loss: 0.1851 - accuracy: 0.9458 - val_loss: 0.2384 - val_accuracy: 0.9463\n",
      "Epoch 12/15\n",
      "529/529 [==============================] - 180s 340ms/step - loss: 0.1856 - accuracy: 0.9458 - val_loss: 0.2397 - val_accuracy: 0.9373\n",
      "Epoch 13/15\n",
      "529/529 [==============================] - 164s 310ms/step - loss: 0.1622 - accuracy: 0.9518 - val_loss: 0.5263 - val_accuracy: 0.8940\n",
      "Epoch 14/15\n",
      "529/529 [==============================] - 162s 306ms/step - loss: 0.1699 - accuracy: 0.9498 - val_loss: 0.3174 - val_accuracy: 0.9219\n",
      "Epoch 15/15\n",
      "529/529 [==============================] - 162s 307ms/step - loss: 0.1569 - accuracy: 0.9534 - val_loss: 0.2133 - val_accuracy: 0.9492\n"
     ]
    }
   ],
   "source": [
    "BATCH_SIZE = 64\n",
    "\n",
    "STEP_PER_EPOCH = len(train_x) // BATCH_SIZE\n",
    "VALIDATION_STEPS = len(test_x) // BATCH_SIZE\n",
    "\n",
    "my_checkpointer = [EarlyStopping(monitor='val_loss', patience=5, verbose=0)]\n",
    "\n",
    "model_history = model.fit(augmentation.flow(train_x, train_y, batch_size=BATCH_SIZE),\n",
    "          steps_per_epoch=STEP_PER_EPOCH,\n",
    "          validation_data=(test_x, test_y),\n",
    "          validation_steps=VALIDATION_STEPS,\n",
    "          epochs=EPOCHS, callbacks=my_checkpointer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "d:\\Github Repo\\monopol\\env\\lib\\site-packages\\keras\\engine\\functional.py:1410: CustomMaskWarning: Custom mask layers require a config and must override get_config. When loading, the custom mask layer must be passed to the custom_objects argument.\n",
      "  layer_config = serialize_layer_fn(layer)\n"
     ]
    }
   ],
   "source": [
    "model.save(\"./src/models/plate_char_classifier_02.h5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x21edeab6220>"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1008x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "figure = plt.figure(figsize=(14,5))\n",
    "grid=gridspec.GridSpec(ncols=2, nrows=1, figure=figure)\n",
    "figure.add_subplot(grid[0])\n",
    "plt.plot(model_history.history['accuracy'], label='training accuracy')\n",
    "plt.plot(model_history.history['val_accuracy'], label='val accuracy')\n",
    "plt.title('Accuracy')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('accuracy')\n",
    "plt.legend()\n",
    "\n",
    "figure.add_subplot(grid[1])\n",
    "plt.plot(model_history.history['loss'], label='training loss')\n",
    "plt.plot(model_history.history['val_loss'], label='val loss')\n",
    "plt.title('Loss')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('loss')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "# save model architectur as json file\n",
    "model_json = model.to_json()\n",
    "with open(\"./src/models/MobileNetV2_plate_character_recognizer.json\", \"w\") as json_file:\n",
    "  json_file.write(model_json)"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "db10861ed731fdbb874e0c00974d8869667160f2f0866ef4e8639eca77d796c7"
  },
  "kernelspec": {
   "display_name": "Python 3.9.7 64-bit ('env': venv)",
   "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": "3.9.7"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}