├── Class Names and Labels
    └── signnames.csv
├── Google Colab Notebook
    └── Traffic_Sign_Classification.ipynb
├── Model
    └── Traffic_Sign_Classifier_CNN.hdf5
├── Procfile
├── README.md
├── Result Excel
    └── Predictions.xlsx
├── Streamlit App Output
    ├── Screenshot (1).png
    ├── Screenshot (2).png
    ├── Screenshot (3).png
    └── streamlit-upload-2020-05-11-13-0.gif
├── Test Random Images
    ├── 80 ends.jpg
    ├── Ahead only.jpg
    ├── Beware_of_Ice_or_snow_Road_Sign.jpg
    ├── Bicycles Crossing.jpg
    ├── Children Crossing.jpg
    ├── Dangerous curve to the left.jpg
    ├── Dangerous curve to the right.jpg
    ├── End of all speed and passing limits.jpg
    ├── End of no passing by vehicles over 3.5 metric tons.jpg
    ├── End of no passing.jpg
    ├── Go straight or left.jpg
    ├── Keep left.jpg
    ├── Keep right.jpg
    ├── No Entry.jpg
    ├── No Passing.jpg
    ├── No Vehicles.jpg
    ├── No passing for vehicles over 3.5 metric tons.jpg
    ├── Pedestrians.jpg
    ├── Road Work.jpg
    ├── Road narrows on the right.jpg
    ├── Roundabout mandatory.jpg
    ├── Traffic signals.jpg
    ├── Turn left ahead.jpg
    ├── Turn right ahead.jpg
    ├── Wild animals crossing.jpg
    ├── bumpy road.jpg
    ├── double_curve.jpg
    ├── general caution.jpg
    ├── go straight or right.jpg
    ├── priority_road.jpg
    ├── right of way at next intersection.jpg
    ├── slippery_road.jpg
    ├── speed limit 100.jpg
    ├── speed_limit_120.jpg
    ├── speed_limit_20.jpg
    ├── speed_limit_30.jpg
    ├── speed_limit_50.jpg
    ├── speed_limit_60.jpg
    ├── speed_limit_70.jpg
    ├── speed_limit_80.jpg
    ├── stop.jpg
    ├── vehicles over 3.5 metric tons prohibited.jpg
    └── yield.jpg
├── app.py
├── classify.py
└── requirements.txt


/Class Names and Labels/signnames.csv:
--------------------------------------------------------------------------------
 1 | ClassId,SignName
 2 | 0,Speed limit (20km/h)
 3 | 1,Speed limit (30km/h)
 4 | 2,Speed limit (50km/h)
 5 | 3,Speed limit (60km/h)
 6 | 4,Speed limit (70km/h)
 7 | 5,Speed limit (80km/h)
 8 | 6,End of speed limit (80km/h)
 9 | 7,Speed limit (100km/h)
10 | 8,Speed limit (120km/h)
11 | 9,No passing
12 | 10,No passing for vehicles over 3.5 metric tons
13 | 11,Right-of-way at the next intersection
14 | 12,Priority road
15 | 13,Yield
16 | 14,Stop
17 | 15,No vehicles
18 | 16,Vehicles over 3.5 metric tons prohibited
19 | 17,No entry
20 | 18,General caution
21 | 19,Dangerous curve to the left
22 | 20,Dangerous curve to the right
23 | 21,Double curve
24 | 22,Bumpy road
25 | 23,Slippery road
26 | 24,Road narrows on the right
27 | 25,Road work
28 | 26,Traffic signals
29 | 27,Pedestrians
30 | 28,Children crossing
31 | 29,Bicycles crossing
32 | 30,Beware of ice/snow
33 | 31,Wild animals crossing
34 | 32,End of all speed and passing limits
35 | 33,Turn right ahead
36 | 34,Turn left ahead
37 | 35,Ahead only
38 | 36,Go straight or right
39 | 37,Go straight or left
40 | 38,Keep right
41 | 39,Keep left
42 | 40,Roundabout mandatory
43 | 41,End of no passing
44 | 42,End of no passing by vehicles over 3.5 metric tons
45 | 


--------------------------------------------------------------------------------
/Google Colab Notebook/Traffic_Sign_Classification.ipynb:
--------------------------------------------------------------------------------
   1 | {
   2 |   "nbformat": 4,
   3 |   "nbformat_minor": 0,
   4 |   "metadata": {
   5 |     "colab": {
   6 |       "name": "Traffic Sign Classification.ipynb",
   7 |       "provenance": [],
   8 |       "collapsed_sections": [],
   9 |       "toc_visible": true
  10 |     },
  11 |     "kernelspec": {
  12 |       "name": "python3",
  13 |       "display_name": "Python 3"
  14 |     },
  15 |     "accelerator": "GPU"
  16 |   },
  17 |   "cells": [
  18 |     {
  19 |       "cell_type": "markdown",
  20 |       "metadata": {
  21 |         "id": "c9JdYuaik6pc",
  22 |         "colab_type": "text"
  23 |       },
  24 |       "source": [
  25 |         "# Import the necessary dependencies "
  26 |       ]
  27 |     },
  28 |     {
  29 |       "cell_type": "markdown",
  30 |       "metadata": {
  31 |         "id": "V08WnIZG446_",
  32 |         "colab_type": "text"
  33 |       },
  34 |       "source": [
  35 |         "### Setting TensorFlow 2.x as default version "
  36 |       ]
  37 |     },
  38 |     {
  39 |       "cell_type": "code",
  40 |       "metadata": {
  41 |         "id": "Lzzsza-ylniV",
  42 |         "colab_type": "code",
  43 |         "colab": {}
  44 |       },
  45 |       "source": [
  46 |         "try:\n",
  47 |         "    %tensorflow_version 2.x\n",
  48 |         "except Exception:\n",
  49 |         "    pass\n",
  50 |         "\n",
  51 |         "import tensorflow as tf"
  52 |       ],
  53 |       "execution_count": 0,
  54 |       "outputs": []
  55 |     },
  56 |     {
  57 |       "cell_type": "markdown",
  58 |       "metadata": {
  59 |         "id": "V8_o8BrZ5Hna",
  60 |         "colab_type": "text"
  61 |       },
  62 |       "source": [
  63 |         "### Importing the Visualisation and Preprocessing packages"
  64 |       ]
  65 |     },
  66 |     {
  67 |       "cell_type": "code",
  68 |       "metadata": {
  69 |         "id": "PM-bq2prNTeq",
  70 |         "colab_type": "code",
  71 |         "colab": {}
  72 |       },
  73 |       "source": [
  74 |         "%matplotlib inline\n",
  75 |         "import matplotlib.pyplot as plt\n",
  76 |         "import matplotlib.image as mpimg\n",
  77 |         "\n",
  78 |         "import os\n",
  79 |         "import cv2\n",
  80 |         "\n",
  81 |         "import numpy as np\n",
  82 |         "import pandas as pd\n",
  83 |         "import pickle\n",
  84 |         "\n",
  85 |         "import random\n"
  86 |       ],
  87 |       "execution_count": 0,
  88 |       "outputs": []
  89 |     },
  90 |     {
  91 |       "cell_type": "markdown",
  92 |       "metadata": {
  93 |         "id": "4s24Ooy07DNB",
  94 |         "colab_type": "text"
  95 |       },
  96 |       "source": [
  97 |         "### Importing the TensorFlow dependencies"
  98 |       ]
  99 |     },
 100 |     {
 101 |       "cell_type": "code",
 102 |       "metadata": {
 103 |         "id": "TvGg5k5oNWIT",
 104 |         "colab_type": "code",
 105 |         "colab": {}
 106 |       },
 107 |       "source": [
 108 |         "from tensorflow.keras.models import Sequential \n",
 109 |         "from tensorflow.keras.layers import Dense,Dropout,Flatten\n",
 110 |         "from tensorflow.keras.layers import BatchNormalization,Conv2D,MaxPooling2D\n",
 111 |         "from tensorflow.keras.utils import to_categorical \n",
 112 |         "from tensorflow.keras.optimizers import Adam\n",
 113 |         "from tensorflow.keras.callbacks import ModelCheckpoint\n",
 114 |         "from tensorflow.keras.regularizers import l2\n",
 115 |         "from tensorflow.keras.preprocessing.image import ImageDataGenerator "
 116 |       ],
 117 |       "execution_count": 0,
 118 |       "outputs": []
 119 |     },
 120 |     {
 121 |       "cell_type": "markdown",
 122 |       "metadata": {
 123 |         "id": "kT2r-te9kyDE",
 124 |         "colab_type": "text"
 125 |       },
 126 |       "source": [
 127 |         "# Data loading and Visualisation "
 128 |       ]
 129 |     },
 130 |     {
 131 |       "cell_type": "markdown",
 132 |       "metadata": {
 133 |         "id": "Rw3x3K3VlXbC",
 134 |         "colab_type": "text"
 135 |       },
 136 |       "source": [
 137 |         "### Loading the data"
 138 |       ]
 139 |     },
 140 |     {
 141 |       "cell_type": "code",
 142 |       "metadata": {
 143 |         "id": "ratYoPLONYND",
 144 |         "colab_type": "code",
 145 |         "outputId": "702bf1e5-b366-47f3-e52c-ecc379ae9ad0",
 146 |         "colab": {
 147 |           "base_uri": "https://localhost:8080/",
 148 |           "height": 72
 149 |         }
 150 |       },
 151 |       "source": [
 152 |         "#data = pd.read_csv(\"german-traffic-signs / signnames.csv\") \n",
 153 |         "  \n",
 154 |         "with open('/content/drive/My Drive/Traffic_sign_Dataset/train.p', 'rb') as f: \n",
 155 |         "    train_data = pickle.load(f) \n",
 156 |         "with open('/content/drive/My Drive/Traffic_sign_Dataset/valid.p', 'rb') as f: \n",
 157 |         "    val_data = pickle.load(f) \n",
 158 |         "with open('/content/drive/My Drive/Traffic_sign_Dataset/test.p', 'rb') as f: \n",
 159 |         "    test_data = pickle.load(f) \n",
 160 |         "\n",
 161 |         "# Extracting the labels from the dictionaries \n",
 162 |         "x_train, y_train = train_data['features'], train_data['labels'] \n",
 163 |         "x_val, y_val = val_data['features'], val_data['labels'] \n",
 164 |         "x_test, y_test = test_data['features'], test_data['labels'] \n",
 165 |         "\n",
 166 |         "print(x_train.shape) \n",
 167 |         "print(x_val.shape) \n",
 168 |         "print(x_test.shape) "
 169 |       ],
 170 |       "execution_count": 0,
 171 |       "outputs": [
 172 |         {
 173 |           "output_type": "stream",
 174 |           "text": [
 175 |             "(34799, 32, 32, 3)\n",
 176 |             "(4410, 32, 32, 3)\n",
 177 |             "(12630, 32, 32, 3)\n"
 178 |           ],
 179 |           "name": "stdout"
 180 |         }
 181 |       ]
 182 |     },
 183 |     {
 184 |       "cell_type": "markdown",
 185 |       "metadata": {
 186 |         "id": "N9Z2-7iellY6",
 187 |         "colab_type": "text"
 188 |       },
 189 |       "source": [
 190 |         "### Visualising the Image data"
 191 |       ]
 192 |     },
 193 |     {
 194 |       "cell_type": "code",
 195 |       "metadata": {
 196 |         "id": "lBhvXPIgNYTB",
 197 |         "colab_type": "code",
 198 |         "outputId": "a137c686-79be-4449-a9db-71f37000a1fe",
 199 |         "colab": {
 200 |           "base_uri": "https://localhost:8080/",
 201 |           "height": 592
 202 |         }
 203 |       },
 204 |       "source": [
 205 |         "import seaborn as sns\n",
 206 |         "import matplotlib.pyplot as plt\n",
 207 |         "\n",
 208 |         "plt.figure(figsize=(15,10))\n",
 209 |         "sns.countplot(x=\"labels\", data=train_data)\n",
 210 |         "sns.despine()"
 211 |       ],
 212 |       "execution_count": 0,
 213 |       "outputs": [
 214 |         {
 215 |           "output_type": "display_data",
 216 |           "data": {
 217 |             "image/png": "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\n",
 218 |             "text/plain": [
 219 |               "<Figure size 1080x720 with 1 Axes>"
 220 |             ]
 221 |           },
 222 |           "metadata": {
 223 |             "tags": [],
 224 |             "needs_background": "light"
 225 |           }
 226 |         }
 227 |       ]
 228 |     },
 229 |     {
 230 |       "cell_type": "markdown",
 231 |       "metadata": {
 232 |         "id": "TfYNP0v54w4I",
 233 |         "colab_type": "text"
 234 |       },
 235 |       "source": [
 236 |         "# Image Processing & Image Augmentation"
 237 |       ]
 238 |     },
 239 |     {
 240 |       "cell_type": "markdown",
 241 |       "metadata": {
 242 |         "id": "AjGV9p6vl47S",
 243 |         "colab_type": "text"
 244 |       },
 245 |       "source": [
 246 |         "### Pre-processing the images using OpenCV"
 247 |       ]
 248 |     },
 249 |     {
 250 |       "cell_type": "code",
 251 |       "metadata": {
 252 |         "id": "mHM2g4w9NcqR",
 253 |         "colab_type": "code",
 254 |         "colab": {}
 255 |       },
 256 |       "source": [
 257 |         "def preprocessing(img): \n",
 258 |         "    img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) \n",
 259 |         "    img = cv2.equalizeHist(img) \n",
 260 |         "    img = img / 255\n",
 261 |         "    return img \n",
 262 |         "  \n",
 263 |         "x_train_final = np.array(list(map(preprocessing, x_train))) \n",
 264 |         "x_val_final = np.array(list(map(preprocessing, x_val))) \n",
 265 |         "x_test_final = np.array(list(map(preprocessing, x_test))) \n",
 266 |         "  \n",
 267 |         "x_train_final = x_train_final.reshape(34799, 32, 32, 1) \n",
 268 |         "x_val_final = x_val_final.reshape(4410, 32, 32, 1) \n",
 269 |         "x_test_final = x_test_final.reshape(12630, 32, 32, 1) \n"
 270 |       ],
 271 |       "execution_count": 0,
 272 |       "outputs": []
 273 |     },
 274 |     {
 275 |       "cell_type": "markdown",
 276 |       "metadata": {
 277 |         "id": "iPxFmAwhmAiB",
 278 |         "colab_type": "text"
 279 |       },
 280 |       "source": [
 281 |         "### Image Augmentation using ImageDataGenerator"
 282 |       ]
 283 |     },
 284 |     {
 285 |       "cell_type": "code",
 286 |       "metadata": {
 287 |         "id": "FzFRYvCXNekZ",
 288 |         "colab_type": "code",
 289 |         "outputId": "514926e7-6de1-4bbe-ab60-d33960620de8",
 290 |         "colab": {
 291 |           "base_uri": "https://localhost:8080/",
 292 |           "height": 192
 293 |         }
 294 |       },
 295 |       "source": [
 296 |         "\n",
 297 |         "datagen = ImageDataGenerator(width_shift_range = 0.1,  \n",
 298 |         "                  height_shift_range = 0.1,  \n",
 299 |         "                  zoom_range = 0.2,  \n",
 300 |         "                  shear_range = 0.1,  \n",
 301 |         "                  rotation_range = 10) \n",
 302 |         "\n",
 303 |         "datagen.fit(x_train_final) \n",
 304 |         "\n",
 305 |         "#os.makedirs('/content/drive/My Drive/Traffic_sign_Dataset/images')\n",
 306 |         "\n",
 307 |         "for x_batch,y_batch in datagen.flow(x_train_final,y_train,batch_size=6,save_to_dir='/content/drive/My Drive/Traffic_sign_Dataset/images',save_prefix='img', save_format='jpg'):\n",
 308 |         "    for i in range(0, 6):\n",
 309 |         "        plt.subplot(330 + 1 + i)\n",
 310 |         "        plt.imshow(x_batch[i].reshape(32, 32), cmap=plt.get_cmap('gray'))\n",
 311 |         "\n",
 312 |         "    plt.show()\n",
 313 |         "    break\n",
 314 |         "\n",
 315 |         "y_train = to_categorical(y_train, 43) \n",
 316 |         "y_val = to_categorical(y_val, 43) \n",
 317 |         "y_test = to_categorical(y_test, 43) \n"
 318 |       ],
 319 |       "execution_count": 0,
 320 |       "outputs": [
 321 |         {
 322 |           "output_type": "display_data",
 323 |           "data": {
 324 |             "image/png": "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\n",
 325 |             "text/plain": [
 326 |               "<Figure size 432x288 with 6 Axes>"
 327 |             ]
 328 |           },
 329 |           "metadata": {
 330 |             "tags": [],
 331 |             "needs_background": "light"
 332 |           }
 333 |         }
 334 |       ]
 335 |     },
 336 |     {
 337 |       "cell_type": "markdown",
 338 |       "metadata": {
 339 |         "id": "2RzYgcur4qyK",
 340 |         "colab_type": "text"
 341 |       },
 342 |       "source": [
 343 |         "# Model Building and Testing "
 344 |       ]
 345 |     },
 346 |     {
 347 |       "cell_type": "markdown",
 348 |       "metadata": {
 349 |         "id": "DYp8FoCtmRO7",
 350 |         "colab_type": "text"
 351 |       },
 352 |       "source": [
 353 |         "### Implementing the CNN Model "
 354 |       ]
 355 |     },
 356 |     {
 357 |       "cell_type": "code",
 358 |       "metadata": {
 359 |         "id": "TAmuD6XvNn9S",
 360 |         "colab_type": "code",
 361 |         "colab": {}
 362 |       },
 363 |       "source": [
 364 |         "def build_model(hp):\n",
 365 |         "    model = Sequential()\n",
 366 |         "\n",
 367 |         "    # Layer 1 \n",
 368 |         "    model.add(Conv2D(filters=hp.Int('CONV_1_FILTER',min_value=32, max_value=64),\n",
 369 |         "                     kernel_size=hp.Choice('KERNEL_1_FILTER', values=[3,5]),\n",
 370 |         "                     activation='relu',\n",
 371 |         "                     input_shape=(32,32,1), \n",
 372 |         "                     padding='same',\n",
 373 |         "                     kernel_regularizer = l2(0.0005)\n",
 374 |         "                     ))\n",
 375 |         "    model.add(MaxPooling2D(pool_size=(2, 2)))\n",
 376 |         "    model.add(Dropout(hp.Float('DROPOUT_1', min_value=0.0,max_value=0.5,default=0.25,step=0.05)))\n",
 377 |         "\n",
 378 |         "    \n",
 379 |         "    # Layer 2 \n",
 380 |         "    model.add(Conv2D(filters=hp.Int('CONV_2_FILTER',min_value=32, max_value=128),\n",
 381 |         "                     kernel_size=hp.Choice('KERNEL_2_FILTER', values=[3,5]),\n",
 382 |         "                     activation='relu',\n",
 383 |         "                     padding='same',\n",
 384 |         "                     kernel_regularizer = l2(0.0005)\n",
 385 |         "                     ))\n",
 386 |         "    model.add(MaxPooling2D(pool_size=(2, 2)))\n",
 387 |         "    model.add(Dropout(hp.Float('DROPOUT_2', min_value=0.0,max_value=0.5,default=0.25,step=0.05)))\n",
 388 |         "\n",
 389 |         "\n",
 390 |         "    # Layer 3\n",
 391 |         "    model.add(Conv2D(filters=hp.Int('CONV_3_FILTER',min_value=32, max_value=128),\n",
 392 |         "                     kernel_size=hp.Choice('KERNEL_3_FILTER', values=[3,5]),\n",
 393 |         "                     activation='relu',\n",
 394 |         "                     padding='same',\n",
 395 |         "                     kernel_regularizer = l2(0.0005)\n",
 396 |         "                     ))\n",
 397 |         "    model.add(MaxPooling2D(pool_size=(2, 2)))\n",
 398 |         "    model.add(Dropout(hp.Float('DROPOUT_3', min_value=0.0,max_value=0.5,default=0.25,step=0.05)))\n",
 399 |         "\n",
 400 |         "    \n",
 401 |         "    model.add(Flatten())\n",
 402 |         "\n",
 403 |         "    \n",
 404 |         "    model.add(Dense(hp.Int('DENSE_1_LAYER', \n",
 405 |         "                           min_value=32, \n",
 406 |         "                           max_value=512),\n",
 407 |         "                    \n",
 408 |         "                    activation = hp.Choice(\n",
 409 |         "                    'dense_activation',\n",
 410 |         "                    values=['relu', 'tanh', 'sigmoid'],\n",
 411 |         "                    default='relu')\n",
 412 |         "    ))\n",
 413 |         "    model.add(Dropout(hp.Float('DROPOUT_2', min_value=0.0,max_value=0.5,default=0.25,step=0.05)))\n",
 414 |         "\n",
 415 |         "    model.add(Dense(43, activation='softmax'))\n",
 416 |         "\n",
 417 |         "    # Learning Rate set between 0.0001 to 0.01\n",
 418 |         "    model.compile(Adam(hp.Float(\n",
 419 |         "                    'learning_rate',\n",
 420 |         "                    min_value=1e-4,\n",
 421 |         "                    max_value=1e-2,\n",
 422 |         "                    sampling='LOG')),\n",
 423 |         "                    loss='categorical_crossentropy', metrics=['accuracy'])\n",
 424 |         "    \n",
 425 |         "    return model\n",
 426 |         "    "
 427 |       ],
 428 |       "execution_count": 0,
 429 |       "outputs": []
 430 |     },
 431 |     {
 432 |       "cell_type": "markdown",
 433 |       "metadata": {
 434 |         "id": "_iAy7irdmXnE",
 435 |         "colab_type": "text"
 436 |       },
 437 |       "source": [
 438 |         "### Install Kerastuner"
 439 |       ]
 440 |     },
 441 |     {
 442 |       "cell_type": "code",
 443 |       "metadata": {
 444 |         "id": "0NOnz_nWej7m",
 445 |         "colab_type": "code",
 446 |         "outputId": "c33dae2f-909d-422d-e486-354a0b1e4bd3",
 447 |         "colab": {
 448 |           "base_uri": "https://localhost:8080/",
 449 |           "height": 313
 450 |         }
 451 |       },
 452 |       "source": [
 453 |         "!pip install keras-tuner"
 454 |       ],
 455 |       "execution_count": 0,
 456 |       "outputs": [
 457 |         {
 458 |           "output_type": "stream",
 459 |           "text": [
 460 |             "Requirement already satisfied: keras-tuner in /usr/local/lib/python3.6/dist-packages (1.0.1)\n",
 461 |             "Requirement already satisfied: scipy in /usr/local/lib/python3.6/dist-packages (from keras-tuner) (1.4.1)\n",
 462 |             "Requirement already satisfied: terminaltables in /usr/local/lib/python3.6/dist-packages (from keras-tuner) (3.1.0)\n",
 463 |             "Requirement already satisfied: requests in /usr/local/lib/python3.6/dist-packages (from keras-tuner) (2.23.0)\n",
 464 |             "Requirement already satisfied: colorama in /usr/local/lib/python3.6/dist-packages (from keras-tuner) (0.4.3)\n",
 465 |             "Requirement already satisfied: scikit-learn in /usr/local/lib/python3.6/dist-packages (from keras-tuner) (0.22.2.post1)\n",
 466 |             "Requirement already satisfied: tabulate in /usr/local/lib/python3.6/dist-packages (from keras-tuner) (0.8.7)\n",
 467 |             "Requirement already satisfied: tqdm in /usr/local/lib/python3.6/dist-packages (from keras-tuner) (4.41.1)\n",
 468 |             "Requirement already satisfied: future in /usr/local/lib/python3.6/dist-packages (from keras-tuner) (0.16.0)\n",
 469 |             "Requirement already satisfied: numpy in /usr/local/lib/python3.6/dist-packages (from keras-tuner) (1.18.4)\n",
 470 |             "Requirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.6/dist-packages (from requests->keras-tuner) (3.0.4)\n",
 471 |             "Requirement already satisfied: urllib3!=1.25.0,!=1.25.1,<1.26,>=1.21.1 in /usr/local/lib/python3.6/dist-packages (from requests->keras-tuner) (1.24.3)\n",
 472 |             "Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.6/dist-packages (from requests->keras-tuner) (2.9)\n",
 473 |             "Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.6/dist-packages (from requests->keras-tuner) (2020.4.5.1)\n",
 474 |             "Requirement already satisfied: joblib>=0.11 in /usr/local/lib/python3.6/dist-packages (from scikit-learn->keras-tuner) (0.14.1)\n"
 475 |           ],
 476 |           "name": "stdout"
 477 |         }
 478 |       ]
 479 |     },
 480 |     {
 481 |       "cell_type": "markdown",
 482 |       "metadata": {
 483 |         "id": "Yt4Sbm9QmbzE",
 484 |         "colab_type": "text"
 485 |       },
 486 |       "source": [
 487 |         "### Use RandomSearch using keraatuner to find the best hyperparameters"
 488 |       ]
 489 |     },
 490 |     {
 491 |       "cell_type": "code",
 492 |       "metadata": {
 493 |         "id": "sf0qjmbaec12",
 494 |         "colab_type": "code",
 495 |         "colab": {}
 496 |       },
 497 |       "source": [
 498 |         "from kerastuner import RandomSearch"
 499 |       ],
 500 |       "execution_count": 0,
 501 |       "outputs": []
 502 |     },
 503 |     {
 504 |       "cell_type": "code",
 505 |       "metadata": {
 506 |         "id": "gsRJmmZheopH",
 507 |         "colab_type": "code",
 508 |         "outputId": "196fd8a5-50df-4640-e1fa-449f3cf4a31c",
 509 |         "colab": {
 510 |           "base_uri": "https://localhost:8080/",
 511 |           "height": 74
 512 |         }
 513 |       },
 514 |       "source": [
 515 |         "tuner = RandomSearch(build_model, objective='val_accuracy',\n",
 516 |         "                     max_trials = 10, \n",
 517 |         "                     directory='/content/drive/My Drive/Traffic_sign_Dataset',\n",
 518 |         "                     project_name=\"Traffic_Sign_Classifier\")"
 519 |       ],
 520 |       "execution_count": 0,
 521 |       "outputs": [
 522 |         {
 523 |           "output_type": "stream",
 524 |           "text": [
 525 |             "INFO:tensorflow:Reloading Oracle from existing project /content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier/oracle.json\n",
 526 |             "INFO:tensorflow:Reloading Tuner from /content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier/tuner0.json\n"
 527 |           ],
 528 |           "name": "stdout"
 529 |         }
 530 |       ]
 531 |     },
 532 |     {
 533 |       "cell_type": "code",
 534 |       "metadata": {
 535 |         "id": "vyB3qo7Nh15E",
 536 |         "colab_type": "code",
 537 |         "outputId": "11eeefa3-c1b8-4d7e-a0ef-f89990956b16",
 538 |         "colab": {
 539 |           "base_uri": "https://localhost:8080/",
 540 |           "height": 1000
 541 |         }
 542 |       },
 543 |       "source": [
 544 |         "tuner.search_space_summary()"
 545 |       ],
 546 |       "execution_count": 0,
 547 |       "outputs": [
 548 |         {
 549 |           "output_type": "display_data",
 550 |           "data": {
 551 |             "text/html": [
 552 |               "<span style=\"color:#4527A0\"><h1 style=\"font-size:18px\">Search space summary</h1></span>"
 553 |             ],
 554 |             "text/plain": [
 555 |               "<IPython.core.display.HTML object>"
 556 |             ]
 557 |           },
 558 |           "metadata": {
 559 |             "tags": []
 560 |           }
 561 |         },
 562 |         {
 563 |           "output_type": "display_data",
 564 |           "data": {
 565 |             "text/html": [
 566 |               "<span style=\"color:cyan\"> |-Default search space size: 12</span>"
 567 |             ],
 568 |             "text/plain": [
 569 |               "<IPython.core.display.HTML object>"
 570 |             ]
 571 |           },
 572 |           "metadata": {
 573 |             "tags": []
 574 |           }
 575 |         },
 576 |         {
 577 |           "output_type": "display_data",
 578 |           "data": {
 579 |             "text/html": [
 580 |               "<span style=\"color:#7E57C2\"><h2 style=\"font-size:16px\">CONV_1_FILTER (Int)</h2></span>"
 581 |             ],
 582 |             "text/plain": [
 583 |               "<IPython.core.display.HTML object>"
 584 |             ]
 585 |           },
 586 |           "metadata": {
 587 |             "tags": []
 588 |           }
 589 |         },
 590 |         {
 591 |           "output_type": "display_data",
 592 |           "data": {
 593 |             "text/html": [
 594 |               "<span style=\"color:cyan\"> |-default: None</span>"
 595 |             ],
 596 |             "text/plain": [
 597 |               "<IPython.core.display.HTML object>"
 598 |             ]
 599 |           },
 600 |           "metadata": {
 601 |             "tags": []
 602 |           }
 603 |         },
 604 |         {
 605 |           "output_type": "display_data",
 606 |           "data": {
 607 |             "text/html": [
 608 |               "<span style=\"color:blue\"> |-max_value: 64</span>"
 609 |             ],
 610 |             "text/plain": [
 611 |               "<IPython.core.display.HTML object>"
 612 |             ]
 613 |           },
 614 |           "metadata": {
 615 |             "tags": []
 616 |           }
 617 |         },
 618 |         {
 619 |           "output_type": "display_data",
 620 |           "data": {
 621 |             "text/html": [
 622 |               "<span style=\"color:cyan\"> |-min_value: 32</span>"
 623 |             ],
 624 |             "text/plain": [
 625 |               "<IPython.core.display.HTML object>"
 626 |             ]
 627 |           },
 628 |           "metadata": {
 629 |             "tags": []
 630 |           }
 631 |         },
 632 |         {
 633 |           "output_type": "display_data",
 634 |           "data": {
 635 |             "text/html": [
 636 |               "<span style=\"color:blue\"> |-sampling: None</span>"
 637 |             ],
 638 |             "text/plain": [
 639 |               "<IPython.core.display.HTML object>"
 640 |             ]
 641 |           },
 642 |           "metadata": {
 643 |             "tags": []
 644 |           }
 645 |         },
 646 |         {
 647 |           "output_type": "display_data",
 648 |           "data": {
 649 |             "text/html": [
 650 |               "<span style=\"color:cyan\"> |-step: 1</span>"
 651 |             ],
 652 |             "text/plain": [
 653 |               "<IPython.core.display.HTML object>"
 654 |             ]
 655 |           },
 656 |           "metadata": {
 657 |             "tags": []
 658 |           }
 659 |         },
 660 |         {
 661 |           "output_type": "display_data",
 662 |           "data": {
 663 |             "text/html": [
 664 |               "<span style=\"color:#7E57C2\"><h2 style=\"font-size:16px\">KERNEL_1_FILTER (Choice)</h2></span>"
 665 |             ],
 666 |             "text/plain": [
 667 |               "<IPython.core.display.HTML object>"
 668 |             ]
 669 |           },
 670 |           "metadata": {
 671 |             "tags": []
 672 |           }
 673 |         },
 674 |         {
 675 |           "output_type": "display_data",
 676 |           "data": {
 677 |             "text/html": [
 678 |               "<span style=\"color:cyan\"> |-default: 3</span>"
 679 |             ],
 680 |             "text/plain": [
 681 |               "<IPython.core.display.HTML object>"
 682 |             ]
 683 |           },
 684 |           "metadata": {
 685 |             "tags": []
 686 |           }
 687 |         },
 688 |         {
 689 |           "output_type": "display_data",
 690 |           "data": {
 691 |             "text/html": [
 692 |               "<span style=\"color:blue\"> |-ordered: True</span>"
 693 |             ],
 694 |             "text/plain": [
 695 |               "<IPython.core.display.HTML object>"
 696 |             ]
 697 |           },
 698 |           "metadata": {
 699 |             "tags": []
 700 |           }
 701 |         },
 702 |         {
 703 |           "output_type": "display_data",
 704 |           "data": {
 705 |             "text/html": [
 706 |               "<span style=\"color:cyan\"> |-values: [3, 5]</span>"
 707 |             ],
 708 |             "text/plain": [
 709 |               "<IPython.core.display.HTML object>"
 710 |             ]
 711 |           },
 712 |           "metadata": {
 713 |             "tags": []
 714 |           }
 715 |         },
 716 |         {
 717 |           "output_type": "display_data",
 718 |           "data": {
 719 |             "text/html": [
 720 |               "<span style=\"color:#7E57C2\"><h2 style=\"font-size:16px\">DROPOUT_1 (Float)</h2></span>"
 721 |             ],
 722 |             "text/plain": [
 723 |               "<IPython.core.display.HTML object>"
 724 |             ]
 725 |           },
 726 |           "metadata": {
 727 |             "tags": []
 728 |           }
 729 |         },
 730 |         {
 731 |           "output_type": "display_data",
 732 |           "data": {
 733 |             "text/html": [
 734 |               "<span style=\"color:cyan\"> |-default: 0.25</span>"
 735 |             ],
 736 |             "text/plain": [
 737 |               "<IPython.core.display.HTML object>"
 738 |             ]
 739 |           },
 740 |           "metadata": {
 741 |             "tags": []
 742 |           }
 743 |         },
 744 |         {
 745 |           "output_type": "display_data",
 746 |           "data": {
 747 |             "text/html": [
 748 |               "<span style=\"color:blue\"> |-max_value: 0.5</span>"
 749 |             ],
 750 |             "text/plain": [
 751 |               "<IPython.core.display.HTML object>"
 752 |             ]
 753 |           },
 754 |           "metadata": {
 755 |             "tags": []
 756 |           }
 757 |         },
 758 |         {
 759 |           "output_type": "display_data",
 760 |           "data": {
 761 |             "text/html": [
 762 |               "<span style=\"color:cyan\"> |-min_value: 0.0</span>"
 763 |             ],
 764 |             "text/plain": [
 765 |               "<IPython.core.display.HTML object>"
 766 |             ]
 767 |           },
 768 |           "metadata": {
 769 |             "tags": []
 770 |           }
 771 |         },
 772 |         {
 773 |           "output_type": "display_data",
 774 |           "data": {
 775 |             "text/html": [
 776 |               "<span style=\"color:blue\"> |-sampling: None</span>"
 777 |             ],
 778 |             "text/plain": [
 779 |               "<IPython.core.display.HTML object>"
 780 |             ]
 781 |           },
 782 |           "metadata": {
 783 |             "tags": []
 784 |           }
 785 |         },
 786 |         {
 787 |           "output_type": "display_data",
 788 |           "data": {
 789 |             "text/html": [
 790 |               "<span style=\"color:cyan\"> |-step: 0.05</span>"
 791 |             ],
 792 |             "text/plain": [
 793 |               "<IPython.core.display.HTML object>"
 794 |             ]
 795 |           },
 796 |           "metadata": {
 797 |             "tags": []
 798 |           }
 799 |         },
 800 |         {
 801 |           "output_type": "display_data",
 802 |           "data": {
 803 |             "text/html": [
 804 |               "<span style=\"color:#7E57C2\"><h2 style=\"font-size:16px\">CONV_2_FILTER (Int)</h2></span>"
 805 |             ],
 806 |             "text/plain": [
 807 |               "<IPython.core.display.HTML object>"
 808 |             ]
 809 |           },
 810 |           "metadata": {
 811 |             "tags": []
 812 |           }
 813 |         },
 814 |         {
 815 |           "output_type": "display_data",
 816 |           "data": {
 817 |             "text/html": [
 818 |               "<span style=\"color:cyan\"> |-default: None</span>"
 819 |             ],
 820 |             "text/plain": [
 821 |               "<IPython.core.display.HTML object>"
 822 |             ]
 823 |           },
 824 |           "metadata": {
 825 |             "tags": []
 826 |           }
 827 |         },
 828 |         {
 829 |           "output_type": "display_data",
 830 |           "data": {
 831 |             "text/html": [
 832 |               "<span style=\"color:blue\"> |-max_value: 128</span>"
 833 |             ],
 834 |             "text/plain": [
 835 |               "<IPython.core.display.HTML object>"
 836 |             ]
 837 |           },
 838 |           "metadata": {
 839 |             "tags": []
 840 |           }
 841 |         },
 842 |         {
 843 |           "output_type": "display_data",
 844 |           "data": {
 845 |             "text/html": [
 846 |               "<span style=\"color:cyan\"> |-min_value: 32</span>"
 847 |             ],
 848 |             "text/plain": [
 849 |               "<IPython.core.display.HTML object>"
 850 |             ]
 851 |           },
 852 |           "metadata": {
 853 |             "tags": []
 854 |           }
 855 |         },
 856 |         {
 857 |           "output_type": "display_data",
 858 |           "data": {
 859 |             "text/html": [
 860 |               "<span style=\"color:blue\"> |-sampling: None</span>"
 861 |             ],
 862 |             "text/plain": [
 863 |               "<IPython.core.display.HTML object>"
 864 |             ]
 865 |           },
 866 |           "metadata": {
 867 |             "tags": []
 868 |           }
 869 |         },
 870 |         {
 871 |           "output_type": "display_data",
 872 |           "data": {
 873 |             "text/html": [
 874 |               "<span style=\"color:cyan\"> |-step: 1</span>"
 875 |             ],
 876 |             "text/plain": [
 877 |               "<IPython.core.display.HTML object>"
 878 |             ]
 879 |           },
 880 |           "metadata": {
 881 |             "tags": []
 882 |           }
 883 |         },
 884 |         {
 885 |           "output_type": "display_data",
 886 |           "data": {
 887 |             "text/html": [
 888 |               "<span style=\"color:#7E57C2\"><h2 style=\"font-size:16px\">KERNEL_2_FILTER (Choice)</h2></span>"
 889 |             ],
 890 |             "text/plain": [
 891 |               "<IPython.core.display.HTML object>"
 892 |             ]
 893 |           },
 894 |           "metadata": {
 895 |             "tags": []
 896 |           }
 897 |         },
 898 |         {
 899 |           "output_type": "display_data",
 900 |           "data": {
 901 |             "text/html": [
 902 |               "<span style=\"color:cyan\"> |-default: 3</span>"
 903 |             ],
 904 |             "text/plain": [
 905 |               "<IPython.core.display.HTML object>"
 906 |             ]
 907 |           },
 908 |           "metadata": {
 909 |             "tags": []
 910 |           }
 911 |         },
 912 |         {
 913 |           "output_type": "display_data",
 914 |           "data": {
 915 |             "text/html": [
 916 |               "<span style=\"color:blue\"> |-ordered: True</span>"
 917 |             ],
 918 |             "text/plain": [
 919 |               "<IPython.core.display.HTML object>"
 920 |             ]
 921 |           },
 922 |           "metadata": {
 923 |             "tags": []
 924 |           }
 925 |         },
 926 |         {
 927 |           "output_type": "display_data",
 928 |           "data": {
 929 |             "text/html": [
 930 |               "<span style=\"color:cyan\"> |-values: [3, 5]</span>"
 931 |             ],
 932 |             "text/plain": [
 933 |               "<IPython.core.display.HTML object>"
 934 |             ]
 935 |           },
 936 |           "metadata": {
 937 |             "tags": []
 938 |           }
 939 |         },
 940 |         {
 941 |           "output_type": "display_data",
 942 |           "data": {
 943 |             "text/html": [
 944 |               "<span style=\"color:#7E57C2\"><h2 style=\"font-size:16px\">DROPOUT_2 (Float)</h2></span>"
 945 |             ],
 946 |             "text/plain": [
 947 |               "<IPython.core.display.HTML object>"
 948 |             ]
 949 |           },
 950 |           "metadata": {
 951 |             "tags": []
 952 |           }
 953 |         },
 954 |         {
 955 |           "output_type": "display_data",
 956 |           "data": {
 957 |             "text/html": [
 958 |               "<span style=\"color:cyan\"> |-default: 0.25</span>"
 959 |             ],
 960 |             "text/plain": [
 961 |               "<IPython.core.display.HTML object>"
 962 |             ]
 963 |           },
 964 |           "metadata": {
 965 |             "tags": []
 966 |           }
 967 |         },
 968 |         {
 969 |           "output_type": "display_data",
 970 |           "data": {
 971 |             "text/html": [
 972 |               "<span style=\"color:blue\"> |-max_value: 0.5</span>"
 973 |             ],
 974 |             "text/plain": [
 975 |               "<IPython.core.display.HTML object>"
 976 |             ]
 977 |           },
 978 |           "metadata": {
 979 |             "tags": []
 980 |           }
 981 |         },
 982 |         {
 983 |           "output_type": "display_data",
 984 |           "data": {
 985 |             "text/html": [
 986 |               "<span style=\"color:cyan\"> |-min_value: 0.0</span>"
 987 |             ],
 988 |             "text/plain": [
 989 |               "<IPython.core.display.HTML object>"
 990 |             ]
 991 |           },
 992 |           "metadata": {
 993 |             "tags": []
 994 |           }
 995 |         },
 996 |         {
 997 |           "output_type": "display_data",
 998 |           "data": {
 999 |             "text/html": [
1000 |               "<span style=\"color:blue\"> |-sampling: None</span>"
1001 |             ],
1002 |             "text/plain": [
1003 |               "<IPython.core.display.HTML object>"
1004 |             ]
1005 |           },
1006 |           "metadata": {
1007 |             "tags": []
1008 |           }
1009 |         },
1010 |         {
1011 |           "output_type": "display_data",
1012 |           "data": {
1013 |             "text/html": [
1014 |               "<span style=\"color:cyan\"> |-step: 0.05</span>"
1015 |             ],
1016 |             "text/plain": [
1017 |               "<IPython.core.display.HTML object>"
1018 |             ]
1019 |           },
1020 |           "metadata": {
1021 |             "tags": []
1022 |           }
1023 |         },
1024 |         {
1025 |           "output_type": "display_data",
1026 |           "data": {
1027 |             "text/html": [
1028 |               "<span style=\"color:#7E57C2\"><h2 style=\"font-size:16px\">CONV_3_FILTER (Int)</h2></span>"
1029 |             ],
1030 |             "text/plain": [
1031 |               "<IPython.core.display.HTML object>"
1032 |             ]
1033 |           },
1034 |           "metadata": {
1035 |             "tags": []
1036 |           }
1037 |         },
1038 |         {
1039 |           "output_type": "display_data",
1040 |           "data": {
1041 |             "text/html": [
1042 |               "<span style=\"color:cyan\"> |-default: None</span>"
1043 |             ],
1044 |             "text/plain": [
1045 |               "<IPython.core.display.HTML object>"
1046 |             ]
1047 |           },
1048 |           "metadata": {
1049 |             "tags": []
1050 |           }
1051 |         },
1052 |         {
1053 |           "output_type": "display_data",
1054 |           "data": {
1055 |             "text/html": [
1056 |               "<span style=\"color:blue\"> |-max_value: 128</span>"
1057 |             ],
1058 |             "text/plain": [
1059 |               "<IPython.core.display.HTML object>"
1060 |             ]
1061 |           },
1062 |           "metadata": {
1063 |             "tags": []
1064 |           }
1065 |         },
1066 |         {
1067 |           "output_type": "display_data",
1068 |           "data": {
1069 |             "text/html": [
1070 |               "<span style=\"color:cyan\"> |-min_value: 32</span>"
1071 |             ],
1072 |             "text/plain": [
1073 |               "<IPython.core.display.HTML object>"
1074 |             ]
1075 |           },
1076 |           "metadata": {
1077 |             "tags": []
1078 |           }
1079 |         },
1080 |         {
1081 |           "output_type": "display_data",
1082 |           "data": {
1083 |             "text/html": [
1084 |               "<span style=\"color:blue\"> |-sampling: None</span>"
1085 |             ],
1086 |             "text/plain": [
1087 |               "<IPython.core.display.HTML object>"
1088 |             ]
1089 |           },
1090 |           "metadata": {
1091 |             "tags": []
1092 |           }
1093 |         },
1094 |         {
1095 |           "output_type": "display_data",
1096 |           "data": {
1097 |             "text/html": [
1098 |               "<span style=\"color:cyan\"> |-step: 1</span>"
1099 |             ],
1100 |             "text/plain": [
1101 |               "<IPython.core.display.HTML object>"
1102 |             ]
1103 |           },
1104 |           "metadata": {
1105 |             "tags": []
1106 |           }
1107 |         },
1108 |         {
1109 |           "output_type": "display_data",
1110 |           "data": {
1111 |             "text/html": [
1112 |               "<span style=\"color:#7E57C2\"><h2 style=\"font-size:16px\">KERNEL_3_FILTER (Choice)</h2></span>"
1113 |             ],
1114 |             "text/plain": [
1115 |               "<IPython.core.display.HTML object>"
1116 |             ]
1117 |           },
1118 |           "metadata": {
1119 |             "tags": []
1120 |           }
1121 |         },
1122 |         {
1123 |           "output_type": "display_data",
1124 |           "data": {
1125 |             "text/html": [
1126 |               "<span style=\"color:cyan\"> |-default: 3</span>"
1127 |             ],
1128 |             "text/plain": [
1129 |               "<IPython.core.display.HTML object>"
1130 |             ]
1131 |           },
1132 |           "metadata": {
1133 |             "tags": []
1134 |           }
1135 |         },
1136 |         {
1137 |           "output_type": "display_data",
1138 |           "data": {
1139 |             "text/html": [
1140 |               "<span style=\"color:blue\"> |-ordered: True</span>"
1141 |             ],
1142 |             "text/plain": [
1143 |               "<IPython.core.display.HTML object>"
1144 |             ]
1145 |           },
1146 |           "metadata": {
1147 |             "tags": []
1148 |           }
1149 |         },
1150 |         {
1151 |           "output_type": "display_data",
1152 |           "data": {
1153 |             "text/html": [
1154 |               "<span style=\"color:cyan\"> |-values: [3, 5]</span>"
1155 |             ],
1156 |             "text/plain": [
1157 |               "<IPython.core.display.HTML object>"
1158 |             ]
1159 |           },
1160 |           "metadata": {
1161 |             "tags": []
1162 |           }
1163 |         },
1164 |         {
1165 |           "output_type": "display_data",
1166 |           "data": {
1167 |             "text/html": [
1168 |               "<span style=\"color:#7E57C2\"><h2 style=\"font-size:16px\">DROPOUT_3 (Float)</h2></span>"
1169 |             ],
1170 |             "text/plain": [
1171 |               "<IPython.core.display.HTML object>"
1172 |             ]
1173 |           },
1174 |           "metadata": {
1175 |             "tags": []
1176 |           }
1177 |         },
1178 |         {
1179 |           "output_type": "display_data",
1180 |           "data": {
1181 |             "text/html": [
1182 |               "<span style=\"color:cyan\"> |-default: 0.25</span>"
1183 |             ],
1184 |             "text/plain": [
1185 |               "<IPython.core.display.HTML object>"
1186 |             ]
1187 |           },
1188 |           "metadata": {
1189 |             "tags": []
1190 |           }
1191 |         },
1192 |         {
1193 |           "output_type": "display_data",
1194 |           "data": {
1195 |             "text/html": [
1196 |               "<span style=\"color:blue\"> |-max_value: 0.5</span>"
1197 |             ],
1198 |             "text/plain": [
1199 |               "<IPython.core.display.HTML object>"
1200 |             ]
1201 |           },
1202 |           "metadata": {
1203 |             "tags": []
1204 |           }
1205 |         },
1206 |         {
1207 |           "output_type": "display_data",
1208 |           "data": {
1209 |             "text/html": [
1210 |               "<span style=\"color:cyan\"> |-min_value: 0.0</span>"
1211 |             ],
1212 |             "text/plain": [
1213 |               "<IPython.core.display.HTML object>"
1214 |             ]
1215 |           },
1216 |           "metadata": {
1217 |             "tags": []
1218 |           }
1219 |         },
1220 |         {
1221 |           "output_type": "display_data",
1222 |           "data": {
1223 |             "text/html": [
1224 |               "<span style=\"color:blue\"> |-sampling: None</span>"
1225 |             ],
1226 |             "text/plain": [
1227 |               "<IPython.core.display.HTML object>"
1228 |             ]
1229 |           },
1230 |           "metadata": {
1231 |             "tags": []
1232 |           }
1233 |         },
1234 |         {
1235 |           "output_type": "display_data",
1236 |           "data": {
1237 |             "text/html": [
1238 |               "<span style=\"color:cyan\"> |-step: 0.05</span>"
1239 |             ],
1240 |             "text/plain": [
1241 |               "<IPython.core.display.HTML object>"
1242 |             ]
1243 |           },
1244 |           "metadata": {
1245 |             "tags": []
1246 |           }
1247 |         },
1248 |         {
1249 |           "output_type": "display_data",
1250 |           "data": {
1251 |             "text/html": [
1252 |               "<span style=\"color:#7E57C2\"><h2 style=\"font-size:16px\">DENSE_1_LAYER (Int)</h2></span>"
1253 |             ],
1254 |             "text/plain": [
1255 |               "<IPython.core.display.HTML object>"
1256 |             ]
1257 |           },
1258 |           "metadata": {
1259 |             "tags": []
1260 |           }
1261 |         },
1262 |         {
1263 |           "output_type": "display_data",
1264 |           "data": {
1265 |             "text/html": [
1266 |               "<span style=\"color:cyan\"> |-default: None</span>"
1267 |             ],
1268 |             "text/plain": [
1269 |               "<IPython.core.display.HTML object>"
1270 |             ]
1271 |           },
1272 |           "metadata": {
1273 |             "tags": []
1274 |           }
1275 |         },
1276 |         {
1277 |           "output_type": "display_data",
1278 |           "data": {
1279 |             "text/html": [
1280 |               "<span style=\"color:blue\"> |-max_value: 512</span>"
1281 |             ],
1282 |             "text/plain": [
1283 |               "<IPython.core.display.HTML object>"
1284 |             ]
1285 |           },
1286 |           "metadata": {
1287 |             "tags": []
1288 |           }
1289 |         },
1290 |         {
1291 |           "output_type": "display_data",
1292 |           "data": {
1293 |             "text/html": [
1294 |               "<span style=\"color:cyan\"> |-min_value: 32</span>"
1295 |             ],
1296 |             "text/plain": [
1297 |               "<IPython.core.display.HTML object>"
1298 |             ]
1299 |           },
1300 |           "metadata": {
1301 |             "tags": []
1302 |           }
1303 |         },
1304 |         {
1305 |           "output_type": "display_data",
1306 |           "data": {
1307 |             "text/html": [
1308 |               "<span style=\"color:blue\"> |-sampling: None</span>"
1309 |             ],
1310 |             "text/plain": [
1311 |               "<IPython.core.display.HTML object>"
1312 |             ]
1313 |           },
1314 |           "metadata": {
1315 |             "tags": []
1316 |           }
1317 |         },
1318 |         {
1319 |           "output_type": "display_data",
1320 |           "data": {
1321 |             "text/html": [
1322 |               "<span style=\"color:cyan\"> |-step: 1</span>"
1323 |             ],
1324 |             "text/plain": [
1325 |               "<IPython.core.display.HTML object>"
1326 |             ]
1327 |           },
1328 |           "metadata": {
1329 |             "tags": []
1330 |           }
1331 |         },
1332 |         {
1333 |           "output_type": "display_data",
1334 |           "data": {
1335 |             "text/html": [
1336 |               "<span style=\"color:#7E57C2\"><h2 style=\"font-size:16px\">dense_activation (Choice)</h2></span>"
1337 |             ],
1338 |             "text/plain": [
1339 |               "<IPython.core.display.HTML object>"
1340 |             ]
1341 |           },
1342 |           "metadata": {
1343 |             "tags": []
1344 |           }
1345 |         },
1346 |         {
1347 |           "output_type": "display_data",
1348 |           "data": {
1349 |             "text/html": [
1350 |               "<span style=\"color:cyan\"> |-default: relu</span>"
1351 |             ],
1352 |             "text/plain": [
1353 |               "<IPython.core.display.HTML object>"
1354 |             ]
1355 |           },
1356 |           "metadata": {
1357 |             "tags": []
1358 |           }
1359 |         },
1360 |         {
1361 |           "output_type": "display_data",
1362 |           "data": {
1363 |             "text/html": [
1364 |               "<span style=\"color:blue\"> |-ordered: False</span>"
1365 |             ],
1366 |             "text/plain": [
1367 |               "<IPython.core.display.HTML object>"
1368 |             ]
1369 |           },
1370 |           "metadata": {
1371 |             "tags": []
1372 |           }
1373 |         },
1374 |         {
1375 |           "output_type": "display_data",
1376 |           "data": {
1377 |             "text/html": [
1378 |               "<span style=\"color:cyan\"> |-values: ['relu', 'tanh', 'sigmoid']</span>"
1379 |             ],
1380 |             "text/plain": [
1381 |               "<IPython.core.display.HTML object>"
1382 |             ]
1383 |           },
1384 |           "metadata": {
1385 |             "tags": []
1386 |           }
1387 |         },
1388 |         {
1389 |           "output_type": "display_data",
1390 |           "data": {
1391 |             "text/html": [
1392 |               "<span style=\"color:#7E57C2\"><h2 style=\"font-size:16px\">learning_rate (Float)</h2></span>"
1393 |             ],
1394 |             "text/plain": [
1395 |               "<IPython.core.display.HTML object>"
1396 |             ]
1397 |           },
1398 |           "metadata": {
1399 |             "tags": []
1400 |           }
1401 |         },
1402 |         {
1403 |           "output_type": "display_data",
1404 |           "data": {
1405 |             "text/html": [
1406 |               "<span style=\"color:cyan\"> |-default: 0.0001</span>"
1407 |             ],
1408 |             "text/plain": [
1409 |               "<IPython.core.display.HTML object>"
1410 |             ]
1411 |           },
1412 |           "metadata": {
1413 |             "tags": []
1414 |           }
1415 |         },
1416 |         {
1417 |           "output_type": "display_data",
1418 |           "data": {
1419 |             "text/html": [
1420 |               "<span style=\"color:blue\"> |-max_value: 0.01</span>"
1421 |             ],
1422 |             "text/plain": [
1423 |               "<IPython.core.display.HTML object>"
1424 |             ]
1425 |           },
1426 |           "metadata": {
1427 |             "tags": []
1428 |           }
1429 |         },
1430 |         {
1431 |           "output_type": "display_data",
1432 |           "data": {
1433 |             "text/html": [
1434 |               "<span style=\"color:cyan\"> |-min_value: 0.0001</span>"
1435 |             ],
1436 |             "text/plain": [
1437 |               "<IPython.core.display.HTML object>"
1438 |             ]
1439 |           },
1440 |           "metadata": {
1441 |             "tags": []
1442 |           }
1443 |         },
1444 |         {
1445 |           "output_type": "display_data",
1446 |           "data": {
1447 |             "text/html": [
1448 |               "<span style=\"color:blue\"> |-sampling: log</span>"
1449 |             ],
1450 |             "text/plain": [
1451 |               "<IPython.core.display.HTML object>"
1452 |             ]
1453 |           },
1454 |           "metadata": {
1455 |             "tags": []
1456 |           }
1457 |         },
1458 |         {
1459 |           "output_type": "display_data",
1460 |           "data": {
1461 |             "text/html": [
1462 |               "<span style=\"color:cyan\"> |-step: None</span>"
1463 |             ],
1464 |             "text/plain": [
1465 |               "<IPython.core.display.HTML object>"
1466 |             ]
1467 |           },
1468 |           "metadata": {
1469 |             "tags": []
1470 |           }
1471 |         }
1472 |       ]
1473 |     },
1474 |     {
1475 |       "cell_type": "code",
1476 |       "metadata": {
1477 |         "id": "3Jo_89zTfyWK",
1478 |         "colab_type": "code",
1479 |         "outputId": "c5555bec-56bf-4183-887d-800638f8b6c5",
1480 |         "colab": {
1481 |           "base_uri": "https://localhost:8080/",
1482 |           "height": 35
1483 |         }
1484 |       },
1485 |       "source": [
1486 |         "tuner.search(x_train_final,y_train,epochs=10, validation_data=(x_val_final,y_val))"
1487 |       ],
1488 |       "execution_count": 0,
1489 |       "outputs": [
1490 |         {
1491 |           "output_type": "stream",
1492 |           "text": [
1493 |             "INFO:tensorflow:Oracle triggered exit\n"
1494 |           ],
1495 |           "name": "stdout"
1496 |         }
1497 |       ]
1498 |     },
1499 |     {
1500 |       "cell_type": "markdown",
1501 |       "metadata": {
1502 |         "id": "YXaYxv7ImoB0",
1503 |         "colab_type": "text"
1504 |       },
1505 |       "source": [
1506 |         "### Get the best model"
1507 |       ]
1508 |     },
1509 |     {
1510 |       "cell_type": "code",
1511 |       "metadata": {
1512 |         "id": "TTZsL2SoiPm4",
1513 |         "colab_type": "code",
1514 |         "colab": {}
1515 |       },
1516 |       "source": [
1517 |         "final_model = tuner.get_best_models(num_models=1)[0]"
1518 |       ],
1519 |       "execution_count": 0,
1520 |       "outputs": []
1521 |     },
1522 |     {
1523 |       "cell_type": "code",
1524 |       "metadata": {
1525 |         "id": "x--itczLkUzm",
1526 |         "colab_type": "code",
1527 |         "outputId": "1544c40a-20d0-4827-e3c8-4104b73dc12e",
1528 |         "colab": {
1529 |           "base_uri": "https://localhost:8080/",
1530 |           "height": 641
1531 |         }
1532 |       },
1533 |       "source": [
1534 |         "final_model.summary()"
1535 |       ],
1536 |       "execution_count": 0,
1537 |       "outputs": [
1538 |         {
1539 |           "output_type": "stream",
1540 |           "text": [
1541 |             "Model: \"sequential\"\n",
1542 |             "_________________________________________________________________\n",
1543 |             "Layer (type)                 Output Shape              Param #   \n",
1544 |             "=================================================================\n",
1545 |             "conv2d (Conv2D)              (None, 32, 32, 48)        480       \n",
1546 |             "_________________________________________________________________\n",
1547 |             "max_pooling2d (MaxPooling2D) (None, 16, 16, 48)        0         \n",
1548 |             "_________________________________________________________________\n",
1549 |             "dropout (Dropout)            (None, 16, 16, 48)        0         \n",
1550 |             "_________________________________________________________________\n",
1551 |             "conv2d_1 (Conv2D)            (None, 16, 16, 92)        110492    \n",
1552 |             "_________________________________________________________________\n",
1553 |             "max_pooling2d_1 (MaxPooling2 (None, 8, 8, 92)          0         \n",
1554 |             "_________________________________________________________________\n",
1555 |             "dropout_1 (Dropout)          (None, 8, 8, 92)          0         \n",
1556 |             "_________________________________________________________________\n",
1557 |             "conv2d_2 (Conv2D)            (None, 8, 8, 127)         292227    \n",
1558 |             "_________________________________________________________________\n",
1559 |             "max_pooling2d_2 (MaxPooling2 (None, 4, 4, 127)         0         \n",
1560 |             "_________________________________________________________________\n",
1561 |             "dropout_2 (Dropout)          (None, 4, 4, 127)         0         \n",
1562 |             "_________________________________________________________________\n",
1563 |             "flatten (Flatten)            (None, 2032)              0         \n",
1564 |             "_________________________________________________________________\n",
1565 |             "dense (Dense)                (None, 217)               441161    \n",
1566 |             "_________________________________________________________________\n",
1567 |             "dropout_3 (Dropout)          (None, 217)               0         \n",
1568 |             "_________________________________________________________________\n",
1569 |             "dense_1 (Dense)              (None, 43)                9374      \n",
1570 |             "=================================================================\n",
1571 |             "Total params: 853,734\n",
1572 |             "Trainable params: 853,734\n",
1573 |             "Non-trainable params: 0\n",
1574 |             "_________________________________________________________________\n"
1575 |           ],
1576 |           "name": "stdout"
1577 |         }
1578 |       ]
1579 |     },
1580 |     {
1581 |       "cell_type": "markdown",
1582 |       "metadata": {
1583 |         "id": "E4GNcwTamsYS",
1584 |         "colab_type": "text"
1585 |       },
1586 |       "source": [
1587 |         "### Creating a Model Checkpoint"
1588 |       ]
1589 |     },
1590 |     {
1591 |       "cell_type": "code",
1592 |       "metadata": {
1593 |         "id": "lLQ5vs52k44x",
1594 |         "colab_type": "code",
1595 |         "colab": {}
1596 |       },
1597 |       "source": [
1598 |         "filepath=\"/content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5\"\n",
1599 |         "checkpoint_conv = ModelCheckpoint(filepath, monitor='val_accuracy', verbose=1, save_best_only=True, mode='max')\n",
1600 |         "callbacks_list_conv = [checkpoint_conv]"
1601 |       ],
1602 |       "execution_count": 0,
1603 |       "outputs": []
1604 |     },
1605 |     {
1606 |       "cell_type": "markdown",
1607 |       "metadata": {
1608 |         "id": "lQeDIaB3m_Ld",
1609 |         "colab_type": "text"
1610 |       },
1611 |       "source": [
1612 |         "### Fit the model"
1613 |       ]
1614 |     },
1615 |     {
1616 |       "cell_type": "code",
1617 |       "metadata": {
1618 |         "id": "PI3l1opFHPCG",
1619 |         "colab_type": "code",
1620 |         "colab": {}
1621 |       },
1622 |       "source": [
1623 |         "BATCH_SIZE = 100\n",
1624 |         "EPOCHS = 40"
1625 |       ],
1626 |       "execution_count": 0,
1627 |       "outputs": []
1628 |     },
1629 |     {
1630 |       "cell_type": "code",
1631 |       "metadata": {
1632 |         "id": "FVrscWrJkZfp",
1633 |         "colab_type": "code",
1634 |         "outputId": "5becc57e-f8bb-4b26-adec-fee393a45b41",
1635 |         "colab": {
1636 |           "base_uri": "https://localhost:8080/",
1637 |           "height": 1000
1638 |         }
1639 |       },
1640 |       "source": [
1641 |         "history = final_model.fit(\n",
1642 |         "                            datagen.flow(x_train_final, y_train, batch_size=BATCH_SIZE), \n",
1643 |         "                            steps_per_epoch = int(np.ceil(len(x_train_final)/float(BATCH_SIZE))), \n",
1644 |         "                            epochs = EPOCHS, \n",
1645 |         "                            validation_data =(x_val_final, y_val),\n",
1646 |         "                            shuffle = True,\n",
1647 |         "                            callbacks=callbacks_list_conv\n",
1648 |         "                          )\n"
1649 |       ],
1650 |       "execution_count": 0,
1651 |       "outputs": [
1652 |         {
1653 |           "output_type": "stream",
1654 |           "text": [
1655 |             "Epoch 1/40\n",
1656 |             "348/348 [==============================] - ETA: 0s - loss: 0.8921 - accuracy: 0.7859\n",
1657 |             "Epoch 00001: val_accuracy improved from -inf to 0.96463, saving model to /content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5\n",
1658 |             "348/348 [==============================] - 15s 42ms/step - loss: 0.8921 - accuracy: 0.7859 - val_loss: 0.2930 - val_accuracy: 0.9646\n",
1659 |             "Epoch 2/40\n",
1660 |             "347/348 [============================>.] - ETA: 0s - loss: 0.6854 - accuracy: 0.8429\n",
1661 |             "Epoch 00002: val_accuracy did not improve from 0.96463\n",
1662 |             "348/348 [==============================] - 12s 36ms/step - loss: 0.6854 - accuracy: 0.8429 - val_loss: 0.2885 - val_accuracy: 0.9637\n",
1663 |             "Epoch 3/40\n",
1664 |             "348/348 [==============================] - ETA: 0s - loss: 0.6061 - accuracy: 0.8646\n",
1665 |             "Epoch 00003: val_accuracy improved from 0.96463 to 0.96735, saving model to /content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5\n",
1666 |             "348/348 [==============================] - 14s 41ms/step - loss: 0.6061 - accuracy: 0.8646 - val_loss: 0.2792 - val_accuracy: 0.9673\n",
1667 |             "Epoch 4/40\n",
1668 |             "348/348 [==============================] - ETA: 0s - loss: 0.5596 - accuracy: 0.8811\n",
1669 |             "Epoch 00004: val_accuracy improved from 0.96735 to 0.97438, saving model to /content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5\n",
1670 |             "348/348 [==============================] - 14s 41ms/step - loss: 0.5596 - accuracy: 0.8811 - val_loss: 0.2707 - val_accuracy: 0.9744\n",
1671 |             "Epoch 5/40\n",
1672 |             "348/348 [==============================] - ETA: 0s - loss: 0.5243 - accuracy: 0.8916\n",
1673 |             "Epoch 00005: val_accuracy improved from 0.97438 to 0.97914, saving model to /content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5\n",
1674 |             "348/348 [==============================] - 14s 42ms/step - loss: 0.5243 - accuracy: 0.8916 - val_loss: 0.2604 - val_accuracy: 0.9791\n",
1675 |             "Epoch 6/40\n",
1676 |             "347/348 [============================>.] - ETA: 0s - loss: 0.4975 - accuracy: 0.9009\n",
1677 |             "Epoch 00006: val_accuracy did not improve from 0.97914\n",
1678 |             "348/348 [==============================] - 13s 38ms/step - loss: 0.4972 - accuracy: 0.9011 - val_loss: 0.2525 - val_accuracy: 0.9785\n",
1679 |             "Epoch 7/40\n",
1680 |             "348/348 [==============================] - ETA: 0s - loss: 0.4719 - accuracy: 0.9098\n",
1681 |             "Epoch 00007: val_accuracy improved from 0.97914 to 0.98095, saving model to /content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5\n",
1682 |             "348/348 [==============================] - 14s 40ms/step - loss: 0.4719 - accuracy: 0.9098 - val_loss: 0.2545 - val_accuracy: 0.9810\n",
1683 |             "Epoch 8/40\n",
1684 |             "347/348 [============================>.] - ETA: 0s - loss: 0.4649 - accuracy: 0.9105\n",
1685 |             "Epoch 00008: val_accuracy did not improve from 0.98095\n",
1686 |             "348/348 [==============================] - 12s 36ms/step - loss: 0.4648 - accuracy: 0.9105 - val_loss: 0.2660 - val_accuracy: 0.9730\n",
1687 |             "Epoch 9/40\n",
1688 |             "347/348 [============================>.] - ETA: 0s - loss: 0.4422 - accuracy: 0.9183\n",
1689 |             "Epoch 00009: val_accuracy did not improve from 0.98095\n",
1690 |             "348/348 [==============================] - 12s 36ms/step - loss: 0.4419 - accuracy: 0.9184 - val_loss: 0.2515 - val_accuracy: 0.9766\n",
1691 |             "Epoch 10/40\n",
1692 |             "347/348 [============================>.] - ETA: 0s - loss: 0.4289 - accuracy: 0.9203\n",
1693 |             "Epoch 00010: val_accuracy did not improve from 0.98095\n",
1694 |             "348/348 [==============================] - 12s 36ms/step - loss: 0.4291 - accuracy: 0.9203 - val_loss: 0.2447 - val_accuracy: 0.9789\n",
1695 |             "Epoch 11/40\n",
1696 |             "347/348 [============================>.] - ETA: 0s - loss: 0.4171 - accuracy: 0.9240\n",
1697 |             "Epoch 00011: val_accuracy improved from 0.98095 to 0.98322, saving model to /content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5\n",
1698 |             "348/348 [==============================] - 14s 41ms/step - loss: 0.4171 - accuracy: 0.9240 - val_loss: 0.2319 - val_accuracy: 0.9832\n",
1699 |             "Epoch 12/40\n",
1700 |             "348/348 [==============================] - ETA: 0s - loss: 0.4154 - accuracy: 0.9236\n",
1701 |             "Epoch 00012: val_accuracy did not improve from 0.98322\n",
1702 |             "348/348 [==============================] - 12s 35ms/step - loss: 0.4154 - accuracy: 0.9236 - val_loss: 0.2504 - val_accuracy: 0.9753\n",
1703 |             "Epoch 13/40\n",
1704 |             "348/348 [==============================] - ETA: 0s - loss: 0.3990 - accuracy: 0.9295\n",
1705 |             "Epoch 00013: val_accuracy did not improve from 0.98322\n",
1706 |             "348/348 [==============================] - 12s 35ms/step - loss: 0.3990 - accuracy: 0.9295 - val_loss: 0.2319 - val_accuracy: 0.9830\n",
1707 |             "Epoch 14/40\n",
1708 |             "348/348 [==============================] - ETA: 0s - loss: 0.3945 - accuracy: 0.9309\n",
1709 |             "Epoch 00014: val_accuracy improved from 0.98322 to 0.98503, saving model to /content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5\n",
1710 |             "348/348 [==============================] - 14s 41ms/step - loss: 0.3945 - accuracy: 0.9309 - val_loss: 0.2297 - val_accuracy: 0.9850\n",
1711 |             "Epoch 15/40\n",
1712 |             "348/348 [==============================] - ETA: 0s - loss: 0.3858 - accuracy: 0.9328\n",
1713 |             "Epoch 00015: val_accuracy did not improve from 0.98503\n",
1714 |             "348/348 [==============================] - 12s 35ms/step - loss: 0.3858 - accuracy: 0.9328 - val_loss: 0.2244 - val_accuracy: 0.9837\n",
1715 |             "Epoch 16/40\n",
1716 |             "348/348 [==============================] - ETA: 0s - loss: 0.3733 - accuracy: 0.9358\n",
1717 |             "Epoch 00016: val_accuracy did not improve from 0.98503\n",
1718 |             "348/348 [==============================] - 12s 35ms/step - loss: 0.3733 - accuracy: 0.9358 - val_loss: 0.2214 - val_accuracy: 0.9848\n",
1719 |             "Epoch 17/40\n",
1720 |             "347/348 [============================>.] - ETA: 0s - loss: 0.3769 - accuracy: 0.9349\n",
1721 |             "Epoch 00017: val_accuracy improved from 0.98503 to 0.98571, saving model to /content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5\n",
1722 |             "348/348 [==============================] - 14s 41ms/step - loss: 0.3766 - accuracy: 0.9350 - val_loss: 0.2185 - val_accuracy: 0.9857\n",
1723 |             "Epoch 18/40\n",
1724 |             "348/348 [==============================] - ETA: 0s - loss: 0.3674 - accuracy: 0.9378\n",
1725 |             "Epoch 00018: val_accuracy did not improve from 0.98571\n",
1726 |             "348/348 [==============================] - 13s 37ms/step - loss: 0.3674 - accuracy: 0.9378 - val_loss: 0.2238 - val_accuracy: 0.9857\n",
1727 |             "Epoch 19/40\n",
1728 |             "348/348 [==============================] - ETA: 0s - loss: 0.3619 - accuracy: 0.9392\n",
1729 |             "Epoch 00019: val_accuracy did not improve from 0.98571\n",
1730 |             "348/348 [==============================] - 12s 35ms/step - loss: 0.3619 - accuracy: 0.9392 - val_loss: 0.2215 - val_accuracy: 0.9839\n",
1731 |             "Epoch 20/40\n",
1732 |             "347/348 [============================>.] - ETA: 0s - loss: 0.3533 - accuracy: 0.9406\n",
1733 |             "Epoch 00020: val_accuracy did not improve from 0.98571\n",
1734 |             "348/348 [==============================] - 12s 35ms/step - loss: 0.3539 - accuracy: 0.9405 - val_loss: 0.2162 - val_accuracy: 0.9853\n",
1735 |             "Epoch 21/40\n",
1736 |             "347/348 [============================>.] - ETA: 0s - loss: 0.3485 - accuracy: 0.9433\n",
1737 |             "Epoch 00021: val_accuracy did not improve from 0.98571\n",
1738 |             "348/348 [==============================] - 12s 35ms/step - loss: 0.3484 - accuracy: 0.9433 - val_loss: 0.2167 - val_accuracy: 0.9855\n",
1739 |             "Epoch 22/40\n",
1740 |             "348/348 [==============================] - ETA: 0s - loss: 0.3477 - accuracy: 0.9430\n",
1741 |             "Epoch 00022: val_accuracy improved from 0.98571 to 0.98798, saving model to /content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5\n",
1742 |             "348/348 [==============================] - 14s 40ms/step - loss: 0.3477 - accuracy: 0.9430 - val_loss: 0.2078 - val_accuracy: 0.9880\n",
1743 |             "Epoch 23/40\n",
1744 |             "348/348 [==============================] - ETA: 0s - loss: 0.3442 - accuracy: 0.9436\n",
1745 |             "Epoch 00023: val_accuracy did not improve from 0.98798\n",
1746 |             "348/348 [==============================] - 13s 37ms/step - loss: 0.3442 - accuracy: 0.9436 - val_loss: 0.2105 - val_accuracy: 0.9862\n",
1747 |             "Epoch 24/40\n",
1748 |             "348/348 [==============================] - ETA: 0s - loss: 0.3382 - accuracy: 0.9445\n",
1749 |             "Epoch 00024: val_accuracy did not improve from 0.98798\n",
1750 |             "348/348 [==============================] - 12s 35ms/step - loss: 0.3382 - accuracy: 0.9445 - val_loss: 0.2074 - val_accuracy: 0.9878\n",
1751 |             "Epoch 25/40\n",
1752 |             "348/348 [==============================] - ETA: 0s - loss: 0.3338 - accuracy: 0.9467\n",
1753 |             "Epoch 00025: val_accuracy did not improve from 0.98798\n",
1754 |             "348/348 [==============================] - 12s 35ms/step - loss: 0.3338 - accuracy: 0.9467 - val_loss: 0.2103 - val_accuracy: 0.9848\n",
1755 |             "Epoch 26/40\n",
1756 |             "348/348 [==============================] - ETA: 0s - loss: 0.3313 - accuracy: 0.9470\n",
1757 |             "Epoch 00026: val_accuracy did not improve from 0.98798\n",
1758 |             "348/348 [==============================] - 12s 35ms/step - loss: 0.3313 - accuracy: 0.9470 - val_loss: 0.2072 - val_accuracy: 0.9868\n",
1759 |             "Epoch 27/40\n",
1760 |             "347/348 [============================>.] - ETA: 0s - loss: 0.3268 - accuracy: 0.9486\n",
1761 |             "Epoch 00027: val_accuracy improved from 0.98798 to 0.98844, saving model to /content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5\n",
1762 |             "348/348 [==============================] - 14s 41ms/step - loss: 0.3268 - accuracy: 0.9486 - val_loss: 0.2055 - val_accuracy: 0.9884\n",
1763 |             "Epoch 28/40\n",
1764 |             "348/348 [==============================] - ETA: 0s - loss: 0.3245 - accuracy: 0.9479\n",
1765 |             "Epoch 00028: val_accuracy did not improve from 0.98844\n",
1766 |             "348/348 [==============================] - 12s 35ms/step - loss: 0.3245 - accuracy: 0.9479 - val_loss: 0.2082 - val_accuracy: 0.9853\n",
1767 |             "Epoch 29/40\n",
1768 |             "347/348 [============================>.] - ETA: 0s - loss: 0.3267 - accuracy: 0.9475\n",
1769 |             "Epoch 00029: val_accuracy did not improve from 0.98844\n",
1770 |             "348/348 [==============================] - 12s 35ms/step - loss: 0.3265 - accuracy: 0.9475 - val_loss: 0.2071 - val_accuracy: 0.9853\n",
1771 |             "Epoch 30/40\n",
1772 |             "347/348 [============================>.] - ETA: 0s - loss: 0.3229 - accuracy: 0.9488\n",
1773 |             "Epoch 00030: val_accuracy did not improve from 0.98844\n",
1774 |             "348/348 [==============================] - 12s 35ms/step - loss: 0.3227 - accuracy: 0.9489 - val_loss: 0.2030 - val_accuracy: 0.9875\n",
1775 |             "Epoch 31/40\n",
1776 |             "348/348 [==============================] - ETA: 0s - loss: 0.3157 - accuracy: 0.9509\n",
1777 |             "Epoch 00031: val_accuracy did not improve from 0.98844\n",
1778 |             "348/348 [==============================] - 12s 35ms/step - loss: 0.3157 - accuracy: 0.9509 - val_loss: 0.2080 - val_accuracy: 0.9832\n",
1779 |             "Epoch 32/40\n",
1780 |             "348/348 [==============================] - ETA: 0s - loss: 0.3132 - accuracy: 0.9516\n",
1781 |             "Epoch 00032: val_accuracy did not improve from 0.98844\n",
1782 |             "348/348 [==============================] - 13s 36ms/step - loss: 0.3132 - accuracy: 0.9516 - val_loss: 0.2009 - val_accuracy: 0.9884\n",
1783 |             "Epoch 33/40\n",
1784 |             "348/348 [==============================] - ETA: 0s - loss: 0.3119 - accuracy: 0.9507\n",
1785 |             "Epoch 00033: val_accuracy improved from 0.98844 to 0.99116, saving model to /content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5\n",
1786 |             "348/348 [==============================] - 14s 40ms/step - loss: 0.3119 - accuracy: 0.9507 - val_loss: 0.1913 - val_accuracy: 0.9912\n",
1787 |             "Epoch 34/40\n",
1788 |             "348/348 [==============================] - ETA: 0s - loss: 0.3112 - accuracy: 0.9511\n",
1789 |             "Epoch 00034: val_accuracy did not improve from 0.99116\n",
1790 |             "348/348 [==============================] - 12s 35ms/step - loss: 0.3112 - accuracy: 0.9511 - val_loss: 0.1986 - val_accuracy: 0.9875\n",
1791 |             "Epoch 35/40\n",
1792 |             "348/348 [==============================] - ETA: 0s - loss: 0.3103 - accuracy: 0.9507\n",
1793 |             "Epoch 00035: val_accuracy did not improve from 0.99116\n",
1794 |             "348/348 [==============================] - 12s 36ms/step - loss: 0.3103 - accuracy: 0.9507 - val_loss: 0.2022 - val_accuracy: 0.9864\n",
1795 |             "Epoch 36/40\n",
1796 |             "347/348 [============================>.] - ETA: 0s - loss: 0.3118 - accuracy: 0.9520\n",
1797 |             "Epoch 00036: val_accuracy did not improve from 0.99116\n",
1798 |             "348/348 [==============================] - 12s 36ms/step - loss: 0.3117 - accuracy: 0.9520 - val_loss: 0.1965 - val_accuracy: 0.9862\n",
1799 |             "Epoch 37/40\n",
1800 |             "348/348 [==============================] - ETA: 0s - loss: 0.3066 - accuracy: 0.9522\n",
1801 |             "Epoch 00037: val_accuracy did not improve from 0.99116\n",
1802 |             "348/348 [==============================] - 13s 37ms/step - loss: 0.3066 - accuracy: 0.9522 - val_loss: 0.1990 - val_accuracy: 0.9875\n",
1803 |             "Epoch 38/40\n",
1804 |             "348/348 [==============================] - ETA: 0s - loss: 0.3009 - accuracy: 0.9542\n",
1805 |             "Epoch 00038: val_accuracy did not improve from 0.99116\n",
1806 |             "348/348 [==============================] - 13s 37ms/step - loss: 0.3009 - accuracy: 0.9542 - val_loss: 0.1963 - val_accuracy: 0.9891\n",
1807 |             "Epoch 39/40\n",
1808 |             "348/348 [==============================] - ETA: 0s - loss: 0.3011 - accuracy: 0.9544\n",
1809 |             "Epoch 00039: val_accuracy improved from 0.99116 to 0.99161, saving model to /content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5\n",
1810 |             "348/348 [==============================] - 15s 42ms/step - loss: 0.3011 - accuracy: 0.9544 - val_loss: 0.1890 - val_accuracy: 0.9916\n",
1811 |             "Epoch 40/40\n",
1812 |             "347/348 [============================>.] - ETA: 0s - loss: 0.3025 - accuracy: 0.9536\n",
1813 |             "Epoch 00040: val_accuracy did not improve from 0.99161\n",
1814 |             "348/348 [==============================] - 13s 37ms/step - loss: 0.3025 - accuracy: 0.9537 - val_loss: 0.1948 - val_accuracy: 0.9866\n"
1815 |           ],
1816 |           "name": "stdout"
1817 |         }
1818 |       ]
1819 |     },
1820 |     {
1821 |       "cell_type": "code",
1822 |       "metadata": {
1823 |         "id": "gHNIDh_ScyAH",
1824 |         "colab_type": "code",
1825 |         "outputId": "76ce4903-437f-4e98-c654-e6a57cc6539f",
1826 |         "colab": {
1827 |           "base_uri": "https://localhost:8080/",
1828 |           "height": 499
1829 |         }
1830 |       },
1831 |       "source": [
1832 |         "acc = history.history['accuracy']\n",
1833 |         "val_acc = history.history['val_accuracy']\n",
1834 |         "\n",
1835 |         "loss = history.history['loss']\n",
1836 |         "val_loss = history.history['val_loss']\n",
1837 |         "\n",
1838 |         "epochs_range = range(EPOCHS)\n",
1839 |         "\n",
1840 |         "plt.figure(figsize=(8, 8))\n",
1841 |         "plt.subplot(1, 2, 1)\n",
1842 |         "plt.plot(epochs_range, acc, label='Training Accuracy')\n",
1843 |         "plt.plot(epochs_range, val_acc, label='Validation Accuracy')\n",
1844 |         "plt.legend(loc='lower right')\n",
1845 |         "plt.title('Training and Validation Accuracy')\n",
1846 |         "\n",
1847 |         "plt.subplot(1, 2, 2)\n",
1848 |         "plt.plot(epochs_range, loss, label='Training Loss')\n",
1849 |         "plt.plot(epochs_range, val_loss, label='Validation Loss')\n",
1850 |         "plt.legend(loc='upper right')\n",
1851 |         "plt.title('Training and Validation Loss')\n",
1852 |         "plt.show()"
1853 |       ],
1854 |       "execution_count": 0,
1855 |       "outputs": [
1856 |         {
1857 |           "output_type": "display_data",
1858 |           "data": {
1859 |             "image/png": "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\n",
1860 |             "text/plain": [
1861 |               "<Figure size 576x576 with 2 Axes>"
1862 |             ]
1863 |           },
1864 |           "metadata": {
1865 |             "tags": [],
1866 |             "needs_background": "light"
1867 |           }
1868 |         }
1869 |       ]
1870 |     },
1871 |     {
1872 |       "cell_type": "markdown",
1873 |       "metadata": {
1874 |         "id": "--iQkw_ynDl5",
1875 |         "colab_type": "text"
1876 |       },
1877 |       "source": [
1878 |         "### Load the model's weights and biases for evaluation on Test set"
1879 |       ]
1880 |     },
1881 |     {
1882 |       "cell_type": "code",
1883 |       "metadata": {
1884 |         "id": "BMMCe9iWmva5",
1885 |         "colab_type": "code",
1886 |         "colab": {}
1887 |       },
1888 |       "source": [
1889 |         "final_model.load_weights(\"/content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5\")\n",
1890 |         "final_model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])"
1891 |       ],
1892 |       "execution_count": 0,
1893 |       "outputs": []
1894 |     },
1895 |     {
1896 |       "cell_type": "code",
1897 |       "metadata": {
1898 |         "id": "kZ1wzyHCm00z",
1899 |         "colab_type": "code",
1900 |         "outputId": "6001978d-cc30-4e02-ffb7-5310d09375e4",
1901 |         "colab": {
1902 |           "base_uri": "https://localhost:8080/",
1903 |           "height": 90
1904 |         }
1905 |       },
1906 |       "source": [
1907 |         "Pred_conv = final_model.evaluate(x_test_final, y_test, verbose=0)\n",
1908 |         "print(\"Results on the test data\")\n",
1909 |         "print(\" \")\n",
1910 |         "print(\"%s- %.2f\" % (final_model.metrics_names[0], Pred_conv[0]))\n",
1911 |         "print(\"%s- %.2f\" % (final_model.metrics_names[1], Pred_conv[1]))"
1912 |       ],
1913 |       "execution_count": 0,
1914 |       "outputs": [
1915 |         {
1916 |           "output_type": "stream",
1917 |           "text": [
1918 |             "Results on the test data\n",
1919 |             " \n",
1920 |             "loss- 0.28\n",
1921 |             "accuracy- 0.97\n"
1922 |           ],
1923 |           "name": "stdout"
1924 |         }
1925 |       ]
1926 |     },
1927 |     {
1928 |       "cell_type": "markdown",
1929 |       "metadata": {
1930 |         "id": "vel7XAkunRuw",
1931 |         "colab_type": "text"
1932 |       },
1933 |       "source": [
1934 |         "### Testing the model on the test dataset"
1935 |       ]
1936 |     },
1937 |     {
1938 |       "cell_type": "code",
1939 |       "metadata": {
1940 |         "id": "lfqA1LFem4z5",
1941 |         "colab_type": "code",
1942 |         "outputId": "97859b01-960f-44b9-bde5-665e8706c7e1",
1943 |         "colab": {
1944 |           "base_uri": "https://localhost:8080/",
1945 |           "height": 505
1946 |         }
1947 |       },
1948 |       "source": [
1949 |         "\n",
1950 |         "plt.imshow(x_test[990]) \n",
1951 |         "print(\"Predicted sign: \"+ str( \n",
1952 |         "        final_model.predict(x_test_final[1700].reshape(1, 32, 32, 1))))\n",
1953 |         "\n",
1954 |         "print('---------------------------------------------------------------------------')\n",
1955 |         "\n",
1956 |         "print(\"Predicted Sign\" + str(final_model.predict_classes(x_test_final[990].reshape(1, 32, 32, 1))))\n"
1957 |       ],
1958 |       "execution_count": 0,
1959 |       "outputs": [
1960 |         {
1961 |           "output_type": "stream",
1962 |           "text": [
1963 |             "Predicted sign: [[5.95169203e-09 1.33304857e-05 4.64417589e-08 8.22311179e-08\n",
1964 |             "  1.00793041e-10 1.53776520e-04 9.99824703e-01 1.95149958e-08\n",
1965 |             "  9.73634437e-11 8.61657005e-14 1.12300655e-10 1.48708246e-08\n",
1966 |             "  1.76557158e-10 6.37280367e-11 4.33579648e-12 4.28027941e-14\n",
1967 |             "  4.16811075e-09 4.14768914e-13 1.11137002e-10 2.02223966e-13\n",
1968 |             "  8.41277603e-11 2.09324294e-11 1.93099575e-14 1.83566842e-11\n",
1969 |             "  1.35439901e-12 2.13375024e-08 3.83618638e-15 9.25231853e-12\n",
1970 |             "  1.41590586e-10 1.22658272e-10 1.58779621e-08 3.06943387e-10\n",
1971 |             "  1.85403493e-07 4.32619947e-11 1.45089789e-12 2.15373902e-13\n",
1972 |             "  8.96582461e-12 1.05680924e-11 1.87089766e-09 6.98752445e-11\n",
1973 |             "  7.32631733e-10 4.96747754e-09 7.83890482e-06]]\n",
1974 |             "---------------------------------------------------------------------------\n",
1975 |             "Predicted Sign[0]\n"
1976 |           ],
1977 |           "name": "stdout"
1978 |         },
1979 |         {
1980 |           "output_type": "display_data",
1981 |           "data": {
1982 |             "image/png": "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\n",
1983 |             "text/plain": [
1984 |               "<Figure size 432x288 with 1 Axes>"
1985 |             ]
1986 |           },
1987 |           "metadata": {
1988 |             "tags": [],
1989 |             "needs_background": "light"
1990 |           }
1991 |         }
1992 |       ]
1993 |     },
1994 |     {
1995 |       "cell_type": "markdown",
1996 |       "metadata": {
1997 |         "id": "mrbxLqyovv0f",
1998 |         "colab_type": "text"
1999 |       },
2000 |       "source": [
2001 |         "### Save the model"
2002 |       ]
2003 |     },
2004 |     {
2005 |       "cell_type": "code",
2006 |       "metadata": {
2007 |         "id": "ygfiDJ0pnVxv",
2008 |         "colab_type": "code",
2009 |         "colab": {}
2010 |       },
2011 |       "source": [
2012 |         "final_model.save('/content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5')"
2013 |       ],
2014 |       "execution_count": 0,
2015 |       "outputs": []
2016 |     },
2017 |     {
2018 |       "cell_type": "markdown",
2019 |       "metadata": {
2020 |         "id": "DN7YN0GwvyoA",
2021 |         "colab_type": "text"
2022 |       },
2023 |       "source": [
2024 |         "### Loading the saved model "
2025 |       ]
2026 |     },
2027 |     {
2028 |       "cell_type": "code",
2029 |       "metadata": {
2030 |         "id": "UdkbTnFqncvc",
2031 |         "colab_type": "code",
2032 |         "outputId": "352a767d-5316-4bfc-fa28-c54ac97c7645",
2033 |         "colab": {
2034 |           "base_uri": "https://localhost:8080/",
2035 |           "height": 641
2036 |         }
2037 |       },
2038 |       "source": [
2039 |         "loaded_model = tf.keras.models.load_model('/content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5')\n",
2040 |         "\n",
2041 |         "loaded_model.summary()"
2042 |       ],
2043 |       "execution_count": 0,
2044 |       "outputs": [
2045 |         {
2046 |           "output_type": "stream",
2047 |           "text": [
2048 |             "Model: \"sequential\"\n",
2049 |             "_________________________________________________________________\n",
2050 |             "Layer (type)                 Output Shape              Param #   \n",
2051 |             "=================================================================\n",
2052 |             "conv2d (Conv2D)              (None, 32, 32, 48)        480       \n",
2053 |             "_________________________________________________________________\n",
2054 |             "max_pooling2d (MaxPooling2D) (None, 16, 16, 48)        0         \n",
2055 |             "_________________________________________________________________\n",
2056 |             "dropout (Dropout)            (None, 16, 16, 48)        0         \n",
2057 |             "_________________________________________________________________\n",
2058 |             "conv2d_1 (Conv2D)            (None, 16, 16, 92)        110492    \n",
2059 |             "_________________________________________________________________\n",
2060 |             "max_pooling2d_1 (MaxPooling2 (None, 8, 8, 92)          0         \n",
2061 |             "_________________________________________________________________\n",
2062 |             "dropout_1 (Dropout)          (None, 8, 8, 92)          0         \n",
2063 |             "_________________________________________________________________\n",
2064 |             "conv2d_2 (Conv2D)            (None, 8, 8, 127)         292227    \n",
2065 |             "_________________________________________________________________\n",
2066 |             "max_pooling2d_2 (MaxPooling2 (None, 4, 4, 127)         0         \n",
2067 |             "_________________________________________________________________\n",
2068 |             "dropout_2 (Dropout)          (None, 4, 4, 127)         0         \n",
2069 |             "_________________________________________________________________\n",
2070 |             "flatten (Flatten)            (None, 2032)              0         \n",
2071 |             "_________________________________________________________________\n",
2072 |             "dense (Dense)                (None, 217)               441161    \n",
2073 |             "_________________________________________________________________\n",
2074 |             "dropout_3 (Dropout)          (None, 217)               0         \n",
2075 |             "_________________________________________________________________\n",
2076 |             "dense_1 (Dense)              (None, 43)                9374      \n",
2077 |             "=================================================================\n",
2078 |             "Total params: 853,734\n",
2079 |             "Trainable params: 853,734\n",
2080 |             "Non-trainable params: 0\n",
2081 |             "_________________________________________________________________\n"
2082 |           ],
2083 |           "name": "stdout"
2084 |         }
2085 |       ]
2086 |     },
2087 |     {
2088 |       "cell_type": "markdown",
2089 |       "metadata": {
2090 |         "id": "T02omtWWv22P",
2091 |         "colab_type": "text"
2092 |       },
2093 |       "source": [
2094 |         "### Making predictions with the loaded model "
2095 |       ]
2096 |     },
2097 |     {
2098 |       "cell_type": "code",
2099 |       "metadata": {
2100 |         "id": "eH6japacnc1B",
2101 |         "colab_type": "code",
2102 |         "outputId": "0ced043a-6325-4c4a-f3bc-2219298a144c",
2103 |         "colab": {
2104 |           "base_uri": "https://localhost:8080/",
2105 |           "height": 468
2106 |         }
2107 |       },
2108 |       "source": [
2109 |         "plt.imshow(x_test[1832]) \n",
2110 |         "print(\"Predicted sign: \"+ str( \n",
2111 |         "        loaded_model.predict(x_test_final[1832].reshape(1, 32, 32, 1))))\n",
2112 |         "\n",
2113 |         "print('---------------------------------------------------------------------------')\n",
2114 |         "\n",
2115 |         "print(\"Predicted Sign\" + str(loaded_model.predict_classes(x_test_final[1832].reshape(1, 32, 32, 1))))"
2116 |       ],
2117 |       "execution_count": 0,
2118 |       "outputs": [
2119 |         {
2120 |           "output_type": "stream",
2121 |           "text": [
2122 |             "Predicted sign: [[3.4522889e-09 3.0688585e-05 1.5866694e-04 6.6440302e-04 9.7419296e-08\n",
2123 |             "  9.9902177e-01 2.3465546e-07 1.2172624e-04 3.9090730e-07 6.0086491e-10\n",
2124 |             "  1.9281763e-06 3.9314019e-10 1.6360911e-10 1.6927328e-11 9.1467207e-12\n",
2125 |             "  4.2105062e-09 1.8575383e-09 6.3886010e-14 8.1647151e-13 6.6127317e-13\n",
2126 |             "  1.0216057e-11 8.7105871e-11 6.1265622e-14 1.7995303e-12 1.5983633e-12\n",
2127 |             "  4.2088120e-09 1.9983780e-13 1.0291361e-13 1.2473573e-11 3.9872386e-10\n",
2128 |             "  1.2319692e-08 1.6544118e-09 1.6150135e-11 5.6740823e-10 4.3921977e-11\n",
2129 |             "  3.0515163e-11 2.1255560e-12 2.0931132e-12 7.5120399e-09 3.6280859e-12\n",
2130 |             "  1.9607171e-09 3.6935528e-15 8.9487511e-09]]\n",
2131 |             "---------------------------------------------------------------------------\n",
2132 |             "Predicted Sign[5]\n"
2133 |           ],
2134 |           "name": "stdout"
2135 |         },
2136 |         {
2137 |           "output_type": "display_data",
2138 |           "data": {
2139 |             "image/png": "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\n",
2140 |             "text/plain": [
2141 |               "<Figure size 432x288 with 1 Axes>"
2142 |             ]
2143 |           },
2144 |           "metadata": {
2145 |             "tags": [],
2146 |             "needs_background": "light"
2147 |           }
2148 |         }
2149 |       ]
2150 |     },
2151 |     {
2152 |       "cell_type": "markdown",
2153 |       "metadata": {
2154 |         "id": "qPc0G00_v7wW",
2155 |         "colab_type": "text"
2156 |       },
2157 |       "source": [
2158 |         "### Testing our saved model on random traffic sign images\n",
2159 |         "\n",
2160 |         "\n",
2161 |         "\n"
2162 |       ]
2163 |     },
2164 |     {
2165 |       "cell_type": "code",
2166 |       "metadata": {
2167 |         "id": "29o5hhOLnwCa",
2168 |         "colab_type": "code",
2169 |         "outputId": "5a6bbd45-602c-48dc-ca2f-66f30acece8e",
2170 |         "colab": {
2171 |           "base_uri": "https://localhost:8080/",
2172 |           "height": 35
2173 |         }
2174 |       },
2175 |       "source": [
2176 |         "from tensorflow.keras.models import load_model\n",
2177 |         "\n",
2178 |         "import cv2\n",
2179 |         "\n",
2180 |         "import numpy as np\n",
2181 |         "\n",
2182 |         "final_model = load_model('/content/drive/My Drive/Traffic_sign_Dataset/Traffic_Sign_Classifier_CNN.hdf5')\n",
2183 |         "\n",
2184 |         "img = cv2.imread('/content/drive/My Drive/Traffic_sign_Dataset/Test Random Images/go straight or right.png')\n",
2185 |         "\n",
2186 |         "img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) \n",
2187 |         "\n",
2188 |         "img = img / 255\n",
2189 |         "\n",
2190 |         "img = cv2.resize(img,(32,32))\n",
2191 |         "\n",
2192 |         "img = np.reshape(img,[1,32,32,1])\n",
2193 |         "\n",
2194 |         "classes = final_model.predict_classes(img)\n",
2195 |         "\n",
2196 |         "print(classes)"
2197 |       ],
2198 |       "execution_count": 0,
2199 |       "outputs": [
2200 |         {
2201 |           "output_type": "stream",
2202 |           "text": [
2203 |             "[36]\n"
2204 |           ],
2205 |           "name": "stdout"
2206 |         }
2207 |       ]
2208 |     },
2209 |     {
2210 |       "cell_type": "code",
2211 |       "metadata": {
2212 |         "id": "hBPbatbDRj4x",
2213 |         "colab_type": "code",
2214 |         "colab": {}
2215 |       },
2216 |       "source": [
2217 |         ""
2218 |       ],
2219 |       "execution_count": 0,
2220 |       "outputs": []
2221 |     }
2222 |   ]
2223 | }


--------------------------------------------------------------------------------
/Model/Traffic_Sign_Classifier_CNN.hdf5:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Model/Traffic_Sign_Classifier_CNN.hdf5


--------------------------------------------------------------------------------
/Procfile:
--------------------------------------------------------------------------------
1 | web: streamlit run --server.enableCORS false --server.port $PORT app.py
2 | 


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
 1 | # Traffic_Sign_Classifier-with-Streamlit
 2 | 
 3 | ## App:
 4 | - Link: https://traffic-sign-classification.herokuapp.com/
 5 | 
 6 | 
 7 | 
 8 | ## App Output
 9 | ![alt text](https://github.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/blob/master/Streamlit%20App%20Output/streamlit-upload-2020-05-11-13-0.gif)
10 | 
11 | 
12 | ## Project overview
13 | - The aim of this project is to focus on the first fundamental features of the decision making ability of an autonomous vehicle, 
14 | i.e., to develop a deep learning model that reads traffic signs and classifies them correctly using Convolutional Neural Networks(CNNs).
15 | 
16 | - The traffic sign classifier uses a German traffic dataset. The German traffic dataset consists of
17 | 34,799 32*32 pixels colored images that is used for the training dataset, 12,630 images are used
18 | for the testing dataset and 4410 images are used in the validation dataset where each images is a
19 | photo of a traffic sign belonging to one of the 43 classes i.e., traffic sign types.
20 | 
21 | 
22 | ## Dataset:
23 | Train Data: https://drive.google.com/open?id=1ZrJJvIbZ5vUHjyzUGNXGl4sRS7zlU5Db
24 | 
25 | Validation Data: https://drive.google.com/open?id=1bLWaYJZHroOyfPVscVdBjh9atjvHDdFj
26 | 
27 | Test Data: https://drive.google.com/open?id=127Usik6jjD_oBhr5hDojgLARW9XYxWdr
28 | 
29 | ## Folders Description
30 | ### Google Colab Notebook
31 | #### Contains the whole process of building the CNN Model 
32 | - Load the Pickled dataset
33 | - Use Seaborn to visualise the data.
34 | - Preprocess the images using OpenCV.
35 | - Use ImageDataGenerator for image augmentation and help the model generalise it's results.
36 | - build_model() function takes hyperparameter(hp) as input and we start building our CNN model using KerasTuner and then compile our model.
37 | - KerasTuner gives us the best hyperparameter combinations using RandomSearch method.
38 | - We now create a model checkpoint and then fit the model and run it for 40 epochs.
39 | - Now Load the model's weights and biases and evaluate it on our test dataset.
40 | - Save our model in Keras HDF5 format.
41 | - Use the saved model to test on random images.
42 | 
43 | ### Test Random Images
44 | - This contains images from the internet. A total of 43 images belonging to each class.
45 | - Our model will be tested using this unseen data
46 | 
47 | ### Streamlit App Output
48 | - Contains the App's final output 
49 | 
50 | ### Class Names and Labels
51 | - Contains the signnames.csv file
52 | 
53 | ### Result Excel
54 | - Conatins a exccel sheet having the results of our test results on random images from the internet
55 | - Also contains the accuracy of our model on unseen data
56 | - Accuracy on unseen data : 79.06%
57 | ### Model
58 | Contains the saved keras model named
59 | - ###### Traffic_sign_classifier_CNN.hdf5
60 | 
61 | ## Files for our Streamlit App
62 | 
63 | #### classify.py
64 | - get_model(): Loads the saved model into cache using streamlit's "@st.cache" feature.
65 | - predict(): Takes an image as input from the function parameter, preprocesses it and feeds it to the model for results.
66 | 
67 | #### app.py
68 | - Contains the front-end code for the streamlit app.
69 | - Imports the predict() function fetches the result and displays it.
70 | 
71 | #### Procfile
72 | A Procfile is a file which describes how to run your application.
73 | 
74 | #### requirements.txt
75 | This has all the dependencies required to deploy our application on Heroku
76 | 
77 | ## Run this app on your system.
78 | ### Requirements
79 | - Python 3.6+
80 | - NumPy
81 | - Pillow
82 | - TensorFlow 2.x
83 | - Streamlit 
84 | 
85 | ### To run it on your system
86 | - Install all the dependencies
87 | - Clone this repository
88 | - You need the Streamlit App folder to run this application.
89 | - In your Command line/Terminal go to the directory where you have upload.py file then type 
90 | #### streamlit run app.py
91 | 
92 | 
93 | 


--------------------------------------------------------------------------------
/Result Excel/Predictions.xlsx:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Result Excel/Predictions.xlsx


--------------------------------------------------------------------------------
/Streamlit App Output/Screenshot (1).png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Streamlit App Output/Screenshot (1).png


--------------------------------------------------------------------------------
/Streamlit App Output/Screenshot (2).png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Streamlit App Output/Screenshot (2).png


--------------------------------------------------------------------------------
/Streamlit App Output/Screenshot (3).png:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Streamlit App Output/Screenshot (3).png


--------------------------------------------------------------------------------
/Streamlit App Output/streamlit-upload-2020-05-11-13-0.gif:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Streamlit App Output/streamlit-upload-2020-05-11-13-0.gif


--------------------------------------------------------------------------------
/Test Random Images/80 ends.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/80 ends.jpg


--------------------------------------------------------------------------------
/Test Random Images/Ahead only.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Ahead only.jpg


--------------------------------------------------------------------------------
/Test Random Images/Beware_of_Ice_or_snow_Road_Sign.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Beware_of_Ice_or_snow_Road_Sign.jpg


--------------------------------------------------------------------------------
/Test Random Images/Bicycles Crossing.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Bicycles Crossing.jpg


--------------------------------------------------------------------------------
/Test Random Images/Children Crossing.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Children Crossing.jpg


--------------------------------------------------------------------------------
/Test Random Images/Dangerous curve to the left.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Dangerous curve to the left.jpg


--------------------------------------------------------------------------------
/Test Random Images/Dangerous curve to the right.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Dangerous curve to the right.jpg


--------------------------------------------------------------------------------
/Test Random Images/End of all speed and passing limits.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/End of all speed and passing limits.jpg


--------------------------------------------------------------------------------
/Test Random Images/End of no passing by vehicles over 3.5 metric tons.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/End of no passing by vehicles over 3.5 metric tons.jpg


--------------------------------------------------------------------------------
/Test Random Images/End of no passing.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/End of no passing.jpg


--------------------------------------------------------------------------------
/Test Random Images/Go straight or left.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Go straight or left.jpg


--------------------------------------------------------------------------------
/Test Random Images/Keep left.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Keep left.jpg


--------------------------------------------------------------------------------
/Test Random Images/Keep right.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Keep right.jpg


--------------------------------------------------------------------------------
/Test Random Images/No Entry.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/No Entry.jpg


--------------------------------------------------------------------------------
/Test Random Images/No Passing.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/No Passing.jpg


--------------------------------------------------------------------------------
/Test Random Images/No Vehicles.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/No Vehicles.jpg


--------------------------------------------------------------------------------
/Test Random Images/No passing for vehicles over 3.5 metric tons.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/No passing for vehicles over 3.5 metric tons.jpg


--------------------------------------------------------------------------------
/Test Random Images/Pedestrians.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Pedestrians.jpg


--------------------------------------------------------------------------------
/Test Random Images/Road Work.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Road Work.jpg


--------------------------------------------------------------------------------
/Test Random Images/Road narrows on the right.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Road narrows on the right.jpg


--------------------------------------------------------------------------------
/Test Random Images/Roundabout mandatory.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Roundabout mandatory.jpg


--------------------------------------------------------------------------------
/Test Random Images/Traffic signals.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Traffic signals.jpg


--------------------------------------------------------------------------------
/Test Random Images/Turn left ahead.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Turn left ahead.jpg


--------------------------------------------------------------------------------
/Test Random Images/Turn right ahead.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Turn right ahead.jpg


--------------------------------------------------------------------------------
/Test Random Images/Wild animals crossing.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/Wild animals crossing.jpg


--------------------------------------------------------------------------------
/Test Random Images/bumpy road.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/bumpy road.jpg


--------------------------------------------------------------------------------
/Test Random Images/double_curve.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/double_curve.jpg


--------------------------------------------------------------------------------
/Test Random Images/general caution.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/general caution.jpg


--------------------------------------------------------------------------------
/Test Random Images/go straight or right.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/go straight or right.jpg


--------------------------------------------------------------------------------
/Test Random Images/priority_road.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/priority_road.jpg


--------------------------------------------------------------------------------
/Test Random Images/right of way at next intersection.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/right of way at next intersection.jpg


--------------------------------------------------------------------------------
/Test Random Images/slippery_road.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/slippery_road.jpg


--------------------------------------------------------------------------------
/Test Random Images/speed limit 100.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/speed limit 100.jpg


--------------------------------------------------------------------------------
/Test Random Images/speed_limit_120.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/speed_limit_120.jpg


--------------------------------------------------------------------------------
/Test Random Images/speed_limit_20.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/speed_limit_20.jpg


--------------------------------------------------------------------------------
/Test Random Images/speed_limit_30.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/speed_limit_30.jpg


--------------------------------------------------------------------------------
/Test Random Images/speed_limit_50.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/speed_limit_50.jpg


--------------------------------------------------------------------------------
/Test Random Images/speed_limit_60.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/speed_limit_60.jpg


--------------------------------------------------------------------------------
/Test Random Images/speed_limit_70.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/speed_limit_70.jpg


--------------------------------------------------------------------------------
/Test Random Images/speed_limit_80.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/speed_limit_80.jpg


--------------------------------------------------------------------------------
/Test Random Images/stop.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/stop.jpg


--------------------------------------------------------------------------------
/Test Random Images/vehicles over 3.5 metric tons prohibited.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/vehicles over 3.5 metric tons prohibited.jpg


--------------------------------------------------------------------------------
/Test Random Images/yield.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/dD2405/Traffic_Sign_Classifier-with-Streamlit/be837a4f7da60531e863f595ef78449672f2f89d/Test Random Images/yield.jpg


--------------------------------------------------------------------------------
/app.py:
--------------------------------------------------------------------------------
 1 | import streamlit as st 
 2 | from PIL import Image
 3 | import classify 
 4 | import numpy as np
 5 | 
 6 | sign_names = {
 7 |         0: 'Speed limit (20km/h)',
 8 |         1: 'Speed limit (30km/h)',
 9 |         2: 'Speed limit (50km/h)',
10 |         3: 'Speed limit (60km/h)',
11 |         4: 'Speed limit (70km/h)',
12 |         5: 'Speed limit (80km/h)',
13 |         6: 'End of speed limit (80km/h)',
14 |         7: 'Speed limit (100km/h)',
15 |         8: 'Speed limit (120km/h)',
16 |         9: 'No passing',
17 |         10: 'No passing for vehicles over 3.5 metric tons',
18 |         11: 'Right-of-way at the next intersection',
19 |         12: 'Priority road',
20 |         13: 'Yield',
21 |         14: 'Stop',
22 |         15: 'No vehicles',
23 |         16: 'Vehicles over 3.5 metric tons prohibited',
24 |         17: 'No entry',
25 |         18: 'General caution',
26 |         19: 'Dangerous curve to the left',
27 |         20: 'Dangerous curve to the right',
28 |         21: 'Double curve',
29 |         22: 'Bumpy road',
30 |         23: 'Slippery road',
31 |         24: 'Road narrows on the right',
32 |         25: 'Road work',
33 |         26: 'Traffic signals',
34 |         27: 'Pedestrians',
35 |         28: 'Children crossing',
36 |         29: 'Bicycles crossing',
37 |         30: 'Beware of ice/snow',
38 |         31: 'Wild animals crossing',
39 |         32: 'End of all speed and passing limits',
40 |         33: 'Turn right ahead',
41 |         34: 'Turn left ahead',
42 |         35: 'Ahead only',
43 |         36: 'Go straight or right',
44 |         37: 'Go straight or left',
45 |         38: 'Keep right',
46 |         39: 'Keep left',
47 |         40: 'Roundabout mandatory',
48 |         41: 'End of no passing',
49 |         42: 'End of no passing by vehicles over 3.5 metric tons'}
50 | 
51 | st.title("Traffic Sign Classifier")
52 | 
53 | uploaded_file = st.file_uploader("Choose an image...", type="jpg")
54 | if uploaded_file is not None:
55 | 
56 |         image = Image.open(uploaded_file)
57 |         st.image(image, caption='Uploaded Image', use_column_width=True)
58 | 
59 |         st.write("")
60 | 
61 |         if st.button('predict'):
62 |                 st.write("Result...")
63 |                 label = classify.predict(uploaded_file)
64 |                 label = label.item()
65 | 
66 |                 res = sign_names.get(label)
67 |                 st.markdown(res)
68 | 


--------------------------------------------------------------------------------
/classify.py:
--------------------------------------------------------------------------------
 1 | from tensorflow.keras.models import load_model
 2 | from tensorflow.keras.preprocessing.image import img_to_array, load_img
 3 | import numpy as np
 4 | import streamlit as st
 5 | 
 6 | @st.cache(allow_output_mutation=True)
 7 | def get_model():
 8 |         model = load_model('Model/Traffic_Sign_Classifier_CNN.hdf5')
 9 |         print('Model Loaded')
10 |         return model 
11 | 
12 |         
13 | def predict(image):
14 |         loaded_model = get_model()
15 |         image = load_img(image, target_size=(32, 32), color_mode = "grayscale")
16 |         image = img_to_array(image)
17 |         image = image/255.0
18 |         image = np.reshape(image,[1,32,32,1])
19 | 
20 |         classes = loaded_model.predict_classes(image)
21 | 
22 |         return classes
23 | 


--------------------------------------------------------------------------------
/requirements.txt:
--------------------------------------------------------------------------------
 1 | cachetools==4.1.0
 2 | decorator==4.3.0
 3 | grpcio==1.25.0
 4 | gunicorn==20.0.4
 5 | h5py==2.9.0
 6 | Keras==2.3.1
 7 | Keras-Applications==1.0.8
 8 | Keras-Preprocessing==1.1.0
 9 | numpy==1.18.3
10 | opencv-python==4.2.0.34
11 | packaging==20.3
12 | Pillow==7.1.1
13 | requests==2.23.0
14 | requests-file==1.4.3
15 | scikit-learn==0.22.2.post1
16 | scipy==1.4.1
17 | sklearn==0.0
18 | streamlit==0.58.0
19 | tensorflow==2.0.0
20 | 


--------------------------------------------------------------------------------