Tensorflow Model for Captcha.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "7217ee60",
   "metadata": {},
   "source": [
    "# Tensorflow CNN model for alphanumeric CAPTCHAs\n",
    "### Notebook to build a Convolutional Neural Network (CNN) model capable of decoding alphanumeric CAPTCHAs of varying lengths and without considering case sensitivity."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "cd0f8a58",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "import os\n",
    "import datetime\n",
    "import random\n",
    "\n",
    "from tensorflow.keras.preprocessing.image import load_img\n",
    "from tensorflow.keras.preprocessing.image import img_to_array\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "#from tensorflow.keras import optimizers, losses, layers\n",
    "from tensorflow.keras.utils import to_categorical\n",
    "#from tensorflow.keras.optimizers.legacy import RMSprop\n",
    "\n",
    "from keras.models import Sequential, Model\n",
    "from keras.layers import Dense, Dropout, Activation, Flatten\n",
    "from tensorflow.keras.layers import Conv2D, MaxPooling2D, Input"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ea907b1e",
   "metadata": {},
   "source": [
    "\n",
    "## 1. Load image dataset "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "a60aa8d9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Nb of files in folder: 104687 \n",
      "\n",
      "Example Image: hkwx7t.jpg\n",
      "Image has shape: (70, 200, 3)\n",
      "The maximum pixel value used is: 255.0\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Dataset directory\n",
    "######################################\n",
    "base_dir = \"./build_dataset/dataset/\"\n",
    "######################################\n",
    "\n",
    "list_dir = os.listdir(base_dir)\n",
    "print(f\"Nb of files in folder: {len(list_dir)} \\n\")\n",
    "\n",
    "# select Random image\n",
    "random_i = random.randint(0, len(list_dir)-1)\n",
    "print(\"Example Image: %s\" % os.listdir(base_dir)[random_i])\n",
    "\n",
    "# Load the Random image\n",
    "source_image  = load_img(f\"{os.path.join(base_dir, list_dir[random_i])}\")\n",
    "# Convert the image into its numpy array representation\n",
    "source_array = img_to_array(source_image)\n",
    "print(f\"Image has shape: {source_array.shape}\")\n",
    "print(f\"The maximum pixel value used is: {np.max(source_array)}\")\n",
    "\n",
    "# Show the Random image\n",
    "plt.imshow(load_img(f\"{os.path.join(base_dir, list_dir[random_i])}\"))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "28c0fc8d",
   "metadata": {},
   "source": [
    "\n",
    "## 2. Generate input data for the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "596f8c62",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Max label length:  7\n",
      "CAPTCHA characters to predict:  [' ', '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z']\n",
      "X train shape (images):  (104687, 70, 200, 1)\n",
      "Y train shape (labels):  (104687, 7, 37)\n",
      "Y train reshaped:  (104687, 259)\n"
     ]
    }
   ],
   "source": [
    "images_array, labels = [],[]\n",
    "list_dir = os.listdir(base_dir)\n",
    "p = np.random.permutation(len(list_dir))\n",
    "\n",
    "# Randomize file order\n",
    "list_dir_new = list_dir\n",
    "for i,j in enumerate(p):\n",
    "    list_dir_new[i] = list_dir[j]\n",
    "\n",
    "# Load files and labels (=file name)\n",
    "for i,file_image in enumerate(list_dir_new):\n",
    "    # filter on images\n",
    "    if ((\"jpg\" in file_image or \"jpeg\" in file_image or \"png\" in file_image) and os.path.getsize(os.path.join(base_dir, file_image)) > 0):\n",
    "\n",
    "        image = load_img(f\"{os.path.join(base_dir, file_image)}\", color_mode = \"grayscale\") \n",
    "        images_array.append( img_to_array(image) ) # Converts a PIL Image instance to a Numpy array.\n",
    "\n",
    "        labels.append( file_image.split(\".\")[0] ) # Extract the image label (= fila name without extension)\n",
    "\n",
    "x_train = images_array\n",
    "y_train = labels\n",
    "\n",
    "\n",
    "# We normalize all label lengths to the same size\n",
    "max_label_len = max(len(label) for label in y_train)\n",
    "print(\"Max label length: \", max_label_len)\n",
    "for i,y in enumerate(y_train):\n",
    "    diff = max_label_len - len(y)\n",
    "    for j in range(diff):\n",
    "        y_train[i] += ' ' \n",
    "        \n",
    "    \n",
    "# We identify all the possible characters of the CAPTCHA\n",
    "characters = set(char for label in y_train for char in label)\n",
    "characters = sorted(list(characters))\n",
    "print(\"CAPTCHA characters to predict: \", characters)\n",
    "\n",
    "\n",
    "# Dictionnaries to map characters to integers\n",
    "char_to_num = {}\n",
    "for i,char in enumerate(characters):\n",
    "    char_to_num[char] = i\n",
    "\n",
    "num_to_char = {}\n",
    "for i,char in enumerate(characters):\n",
    "    num_to_char[i] = char\n",
    "\n",
    "    \n",
    "x_train_num = np.array(x_train) # convert to numpy array\n",
    "\n",
    "\n",
    "y_train_num = []\n",
    "for y_train_i in y_train:\n",
    " \n",
    "    y_train_new_i = []\n",
    "    for j in range(max_label_len): #\n",
    "        # Convert to vector format\n",
    "        if j < len(y_train_i):\n",
    "            y_train_new_i.append( to_categorical(char_to_num[y_train_i[j]], num_classes = len(characters), dtype =\"int32\") ) \n",
    "        else:\n",
    "            y_train_new_i.append(np.zeros(len(characters), dtype=int))\n",
    "    y_train_num.append(y_train_new_i)\n",
    "y_train_num =  np.array(y_train_num) # convert to numpy array\n",
    "\n",
    "#Display one image to check\n",
    "#check =  y_train[5].strip() + \".jpg\"\n",
    "#plt.imshow(load_img(f\"{os.path.join(base_dir, check)}\"))\n",
    "#plt.show()\n",
    "\n",
    "\n",
    "print(\"X train shape (images): \", x_train_num.shape)\n",
    "print(\"Y train shape (labels): \", y_train_num.shape)\n",
    "\n",
    "# Reshape label vectors to single dimension\n",
    "a,b,c = y_train_num.shape\n",
    "y_train_num = y_train_num.reshape(a,b*c)\n",
    "print(\"Y train reshaped: \", y_train_num.shape)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "094735d4",
   "metadata": {},
   "source": [
    "\n",
    "# 3. Augment data\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "667f8ab9",
   "metadata": {},
   "outputs": [],
   "source": [
    "#########################\n",
    "VALIDATION_SPLIT = 0.005\n",
    "#########################\n",
    "\n",
    "\n",
    "# Data Augmentation (uncomment below when the dataset is too small)\n",
    "train_datagen = ImageDataGenerator(rescale = 1./255.\n",
    "                                   #   , rotation_range = 4\n",
    "                                   #   , width_shift_range =  0.02 # it's actually heigh in our case\n",
    "                                   #   , height_shift_range = 0.08 # it's actually width in our case\n",
    "                                   #   , shear_range = 10  \n",
    "                                   #   , zoom_range = 0.08 \n",
    "                                   , validation_split = VALIDATION_SPLIT\n",
    "                             )\n",
    "\n",
    "# No augmentation on test data https://github.com/keras-team/keras/issues/5862#issuecomment-647559571\n",
    "test_datagen = tf.keras.preprocessing.image.ImageDataGenerator(\n",
    "    rescale=1./255\n",
    "   , validation_split=VALIDATION_SPLIT\n",
    ")    \n",
    "\n",
    "train_generator = train_datagen.flow((x_train_num,  y_train_num)\n",
    "                                     #, seed=42\n",
    "                                     #,  shuffle=True\n",
    "                                     , batch_size=256\n",
    "                                     , subset='training'\n",
    "                                     #, save_to_dir='./augmentation/'\n",
    "                                    )\n",
    "\n",
    "test_generator = test_datagen.flow((x_train_num , y_train_num)\n",
    "                                   #, seed=42\n",
    "                                   #, shuffle=True\n",
    "                                   , batch_size=16\n",
    "                                   , subset='validation'\n",
    "                                   )"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "55538269",
   "metadata": {},
   "source": [
    "\n",
    "## 4. Train model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "a0f18e24",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input_1 (InputLayer)        [(None, 70, 200, 1)]      0         \n",
      "                                                                 \n",
      " conv2d (Conv2D)             (None, 70, 200, 32)       832       \n",
      "                                                                 \n",
      " max_pooling2d (MaxPooling2  (None, 35, 100, 32)       0         \n",
      " D)                                                              \n",
      "                                                                 \n",
      " conv2d_1 (Conv2D)           (None, 35, 100, 48)       38448     \n",
      "                                                                 \n",
      " max_pooling2d_1 (MaxPoolin  (None, 17, 50, 48)        0         \n",
      " g2D)                                                            \n",
      "                                                                 \n",
      " conv2d_2 (Conv2D)           (None, 17, 50, 64)        76864     \n",
      "                                                                 \n",
      " max_pooling2d_2 (MaxPoolin  (None, 8, 25, 64)         0         \n",
      " g2D)                                                            \n",
      "                                                                 \n",
      " flatten (Flatten)           (None, 12800)             0         \n",
      "                                                                 \n",
      " dense (Dense)               (None, 259)               3315459   \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 3431603 (13.09 MB)\n",
      "Trainable params: 3431603 (13.09 MB)\n",
      "Non-trainable params: 0 (0.00 Byte)\n",
      "_________________________________________________________________\n",
      "None\n",
      "Epoch 1/20\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2023-08-27 11:58:30.165347: I tensorflow/core/common_runtime/executor.cc:1210] [/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",
      "2023-08-27 12:12:24.983667: I tensorflow/core/common_runtime/executor.cc:1210] [/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": [
      "407/407 - 836s - loss: 0.1029 - accuracy: 0.0098 - val_loss: 0.0560 - val_accuracy: 0.0918 - 836s/epoch - 2s/step\n",
      "Epoch 2/20\n",
      "407/407 - 884s - loss: 0.0287 - accuracy: 0.1832 - val_loss: 0.0120 - val_accuracy: 0.3212 - 884s/epoch - 2s/step\n",
      "Epoch 3/20\n",
      "407/407 - 863s - loss: 0.0092 - accuracy: 0.2655 - val_loss: 0.0061 - val_accuracy: 0.2772 - 863s/epoch - 2s/step\n",
      "Epoch 4/20\n",
      "407/407 - 872s - loss: 0.0047 - accuracy: 0.2555 - val_loss: 0.0035 - val_accuracy: 0.2314 - 872s/epoch - 2s/step\n",
      "Epoch 5/20\n",
      "407/407 - 890s - loss: 0.0028 - accuracy: 0.2229 - val_loss: 0.0022 - val_accuracy: 0.1453 - 890s/epoch - 2s/step\n",
      "Epoch 6/20\n",
      "407/407 - 886s - loss: 0.0019 - accuracy: 0.1794 - val_loss: 0.0016 - val_accuracy: 0.1319 - 886s/epoch - 2s/step\n",
      "Epoch 7/20\n",
      "407/407 - 888s - loss: 0.0012 - accuracy: 0.1472 - val_loss: 0.0011 - val_accuracy: 0.1300 - 888s/epoch - 2s/step\n",
      "Epoch 8/20\n",
      "407/407 - 877s - loss: 9.3297e-04 - accuracy: 0.1194 - val_loss: 9.6084e-04 - val_accuracy: 0.1071 - 877s/epoch - 2s/step\n",
      "Epoch 9/20\n",
      "407/407 - 888s - loss: 7.3024e-04 - accuracy: 0.1039 - val_loss: 8.2276e-04 - val_accuracy: 0.0765 - 888s/epoch - 2s/step\n",
      "Epoch 10/20\n",
      "407/407 - 887s - loss: 6.3428e-04 - accuracy: 0.0795 - val_loss: 7.1010e-04 - val_accuracy: 0.1013 - 887s/epoch - 2s/step\n",
      "Epoch 11/20\n",
      "407/407 - 887s - loss: 6.4102e-04 - accuracy: 0.0845 - val_loss: 6.9176e-04 - val_accuracy: 0.0650 - 887s/epoch - 2s/step\n",
      "Epoch 12/20\n",
      "407/407 - 880s - loss: 6.0425e-04 - accuracy: 0.0855 - val_loss: 6.4328e-04 - val_accuracy: 0.0593 - 880s/epoch - 2s/step\n",
      "Epoch 13/20\n",
      "407/407 - 906s - loss: 4.1955e-04 - accuracy: 0.0730 - val_loss: 3.6718e-04 - val_accuracy: 0.0268 - 906s/epoch - 2s/step\n",
      "Epoch 14/20\n",
      "407/407 - 908s - loss: 4.6099e-04 - accuracy: 0.0659 - val_loss: 3.7791e-04 - val_accuracy: 0.0707 - 908s/epoch - 2s/step\n",
      "Epoch 15/20\n",
      "407/407 - 898s - loss: 3.8716e-04 - accuracy: 0.0785 - val_loss: 5.2132e-04 - val_accuracy: 0.0707 - 898s/epoch - 2s/step\n",
      "Epoch 16/20\n",
      "407/407 - 894s - loss: 3.6353e-04 - accuracy: 0.0758 - val_loss: 2.9012e-04 - val_accuracy: 0.0593 - 894s/epoch - 2s/step\n",
      "Epoch 17/20\n",
      "407/407 - 904s - loss: 3.9607e-04 - accuracy: 0.0934 - val_loss: 4.0477e-04 - val_accuracy: 0.0822 - 904s/epoch - 2s/step\n",
      "Epoch 18/20\n",
      "407/407 - 908s - loss: 3.3569e-04 - accuracy: 0.0996 - val_loss: 3.5695e-04 - val_accuracy: 0.0860 - 908s/epoch - 2s/step\n",
      "Epoch 19/20\n",
      "407/407 - 904s - loss: 3.4380e-04 - accuracy: 0.0752 - val_loss: 2.9700e-04 - val_accuracy: 0.0535 - 904s/epoch - 2s/step\n",
      "Epoch 20/20\n",
      "407/407 - 903s - loss: 2.9677e-04 - accuracy: 0.0804 - val_loss: 4.5288e-04 - val_accuracy: 0.0650 - 903s/epoch - 2s/step\n"
     ]
    }
   ],
   "source": [
    "##############\n",
    "EPOCHS = 20\n",
    "##############\n",
    "\n",
    "class myCallback(tf.keras.callbacks.Callback):\n",
    "    def on_epoch_end(self, epoch, logs={}):\n",
    "        if (logs.get('accuracy') is not None and logs.get('accuracy') > 0.60) and( logs.get('val_accuracy') is not None and logs.get('val_accuracy') >= 0.85):\n",
    "            print(\"\\nReached 60% accuracy and val_accuracy > 85% so stopping training!\")\n",
    "            self.model.stop_training = True    \n",
    "\n",
    "            \n",
    "# Instantiate the callback\n",
    "callbacks = myCallback()\n",
    "\n",
    "\n",
    "input_layer = Input((70, 200, 1))\n",
    "x = Conv2D(filters=32, kernel_size=(5, 5), padding='same', activation='relu')(input_layer)\n",
    "x = MaxPooling2D(pool_size=(2, 2))(x)\n",
    "x = Conv2D(filters=48, kernel_size=(5, 5), padding='same', activation='relu')(x)\n",
    "x = MaxPooling2D(pool_size=(2, 2))(x)\n",
    "x = Conv2D(filters=64, kernel_size=(5, 5), padding='same', activation='relu')(x)\n",
    "x = MaxPooling2D(pool_size=(2, 2))(x)\n",
    "#x = Conv2D(filters=64, kernel_size=(5, 5), padding='same', activation='relu')(x)\n",
    "#x = MaxPooling2D(pool_size=(2, 2))(x)\n",
    "#x = Dropout(0.3)(x)\n",
    "x = Flatten()(x)\n",
    "#x = Dense(1024, activation='relu')(x)\n",
    "#x = Dropout(0.3)(x)\n",
    "#x = Dense(435, activation='relu')(x) \n",
    "#x = Dropout(0.3)(x)\n",
    "\n",
    "#out = [Dense(len(characters), name='digit%d' % i, activation='softmax')(x) for i in range(max_label_len)]\n",
    "out = Dense(len(characters)*max_label_len, activation='sigmoid')(x)\n",
    "\n",
    "model = Model(inputs=input_layer, outputs=out)\n",
    "\n",
    "\n",
    "\n",
    "# Compile the model\n",
    "model.compile(loss='binary_crossentropy', \n",
    "              optimizer='adam',\n",
    "              metrics=['accuracy'])     \n",
    "\n",
    "print(model.summary())\n",
    "\n",
    "\n",
    "# Train the model\n",
    "history = model.fit( train_generator\n",
    "                    , validation_data=test_generator\n",
    "                    , epochs=EPOCHS\n",
    "                    , callbacks=[callbacks]\n",
    "                    , verbose=2\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ebbe487f",
   "metadata": {},
   "source": [
    "\n",
    "## 5. Plot the results for both the training and test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "b2aeff33",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#-----------------------------------------------------------\n",
    "# Retrieve a list of list results on training and test data\n",
    "# sets for each training epoch\n",
    "#-----------------------------------------------------------\n",
    "acc=history.history['accuracy']\n",
    "val_acc=history.history['val_accuracy']\n",
    "loss=history.history['loss']\n",
    "val_loss=history.history['val_loss']\n",
    "\n",
    "epochs=range(len(acc)) # Get number of epochs\n",
    "\n",
    "#------------------------------------------------\n",
    "# Plot training and validation accuracy per epoch\n",
    "#------------------------------------------------\n",
    "plt.plot(epochs, acc, 'r', \"Training Accuracy\")\n",
    "plt.plot(epochs, val_acc, 'b', \"Validation Accuracy\")\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.show()\n",
    "print(\"\")\n",
    "\n",
    "#------------------------------------------------\n",
    "# Plot training and validation loss per epoch\n",
    "#------------------------------------------------\n",
    "plt.plot(epochs, loss, 'r', \"Training Loss\")\n",
    "plt.plot(epochs, val_loss, 'b', \"Validation Loss\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c375ea78",
   "metadata": {},
   "source": [
    "## 6. Check the predictions for a few images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "ef48c8dc",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 98ms/step\n",
      "\n",
      "6madmo.jpg\n",
      "Predicted shape: (1, 70, 200, 1)\n",
      "6madmo\n",
      "Total score : 99.8%\n",
      "Char. scores: 1.0 1.0 1.0 1.0 1.0 1.0 1.0 \n",
      "1/1 [==============================] - 0s 13ms/step\n",
      "\n",
      "x3dspk.jpg\n",
      "Predicted shape: (1, 70, 200, 1)\n",
      "x3dspk\n",
      "Total score : 100.0%\n",
      "Char. scores: 1.0 1.0 1.0 1.0 1.0 1.0 1.0 \n",
      "1/1 [==============================] - 0s 12ms/step\n",
      "\n",
      "sbxlzs.jpg\n",
      "Predicted shape: (1, 70, 200, 1)\n",
      "sbxlzs\n",
      "Total score : 100.0%\n",
      "Char. scores: 1.0 1.0 1.0 1.0 1.0 1.0 1.0 \n",
      "1/1 [==============================] - 0s 12ms/step\n",
      "\n",
      "gtglaf.jpg\n",
      "Predicted shape: (1, 70, 200, 1)\n",
      "gtglaf\n",
      "Total score : 0.2%\n",
      "Char. scores: 1.0 0.0 1.0 1.0 1.0 1.0 1.0 \n"
     ]
    }
   ],
   "source": [
    "check_files = [list_dir_new[0], list_dir_new[1], list_dir_new[2], list_dir_new[3]]\n",
    "\n",
    "for file_i in check_files:\n",
    "    test_image  = load_img(base_dir + file_i, color_mode = \"grayscale\")\n",
    "    test_array = img_to_array(test_image)  # Converts a PIL Image instance to a Numpy array.\n",
    "    test_array /=  255.0 # Convert to decimal value\n",
    "    test_array = np.expand_dims(test_array, axis=0) # add a dimension\n",
    "    result = model.predict(test_array)\n",
    "    result = result.reshape(1, max_label_len, len(characters)) # reshape predicted label to multi verctor dimensions\n",
    "    print(f\"\\n{file_i}\")\n",
    "    print(\"Predicted shape: %s\" % str(test_array.shape))\n",
    "    text = \"\"\n",
    "    text_check = \"Char. scores: \"\n",
    "    text_score = 1\n",
    "    for i in range(max_label_len):\n",
    "        text += num_to_char[np.argmax(result[0][i])] # retrieve the position of the biggest predicted value for each vector\n",
    "        text_check += str(round(result[0][i][np.argmax(result[0][i])], 2)) + \" \" \n",
    "        text_score *= result[0][i][np.argmax(result[0][i])]\n",
    "    text = text.strip() # remove end spaces\n",
    "    print(text)\n",
    "    print(\"Total score : {:.1%}\".format(text_score))\n",
    "    print(text_check)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9b3182ac",
   "metadata": {},
   "source": [
    "\n",
    "## 6. Save model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "81cd59b4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ./models/model_2023-08-27_16h54.pkl/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: ./models/model_2023-08-27_16h54.pkl/assets\n"
     ]
    }
   ],
   "source": [
    "save_dir = \"./models/\"\n",
    "\n",
    "def save_history():\n",
    "  today = datetime.datetime.today() \n",
    "  today_str = today.strftime('%Y-%m-%d_%Hh%M')\n",
    "  file_name = save_dir + 'model_%s.pkl' % today_str\n",
    "    \n",
    "  model.save(file_name)\n",
    "    \n",
    "save_history()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1b1989ef",
   "metadata": {},
   "source": [
    "\n",
    "## 7. Evaluate the model on the whole dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "9f208909",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "818/818 [==============================] - 266s 325ms/step - loss: 16.5286 - accuracy: 0.3588\n",
      "test loss, test accuracy: [16.52863121032715, 0.35879334807395935]\n"
     ]
    }
   ],
   "source": [
    "\n",
    "results = model.evaluate(x_train_num, y_train_num, batch_size=128)\n",
    "print(\"test loss, test accuracy:\", results)\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.16"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}