traffic-sign-classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Traffic Sign Classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Problem Statement\n",
    "\n",
    "Traffic Sign Classification. This project is useful for all autonomous vehicles. And able to identify traffic signs from the image."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dataset Description:\n",
    "\n",
    "The dataset contains more than 50,000 images of different traffic signs. The size of the dataset is around 600 MB. The dataset has a train folder which contains images inside each class and a test folder which we will use for testing our model."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Dataset link: https://www.kaggle.com/datasets/meowmeowmeowmeowmeow/gtsrb-german-traffic-sign"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Methodology to solve the problem\n",
    "\n",
    "1. Explore the dataset\n",
    "2. Splitting the data into Train and Test Set\n",
    "3. Build a CNN model\n",
    "4. Train and validate the model\n",
    "5. Test the model with test dataset\n",
    "6. Predection"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Import Libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-22T08:19:28.732131Z",
     "iopub.status.busy": "2022-04-22T08:19:28.731883Z",
     "iopub.status.idle": "2022-04-22T08:19:28.737290Z",
     "shell.execute_reply": "2022-04-22T08:19:28.736558Z",
     "shell.execute_reply.started": "2022-04-22T08:19:28.732103Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Enabling eager execution\n",
      "INFO:tensorflow:Enabling v2 tensorshape\n",
      "INFO:tensorflow:Enabling resource variables\n",
      "INFO:tensorflow:Enabling tensor equality\n",
      "INFO:tensorflow:Enabling control flow v2\n"
     ]
    }
   ],
   "source": [
    "import numpy as np \n",
    "import pandas as pd \n",
    "import matplotlib.pyplot as plt\n",
    "import cv2\n",
    "import tensorflow as tf\n",
    "from PIL import Image\n",
    "from tqdm import tqdm_notebook as tqdm\n",
    "import os\n",
    "from sklearn.model_selection import train_test_split\n",
    "from tensorflow.keras.utils import to_categorical\n",
    "from keras.models import Sequential, load_model\n",
    "from keras.layers import Conv2D, MaxPool2D, Dense, Flatten, Dropout\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Import Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-22T08:19:28.790849Z",
     "iopub.status.busy": "2022-04-22T08:19:28.790486Z",
     "iopub.status.idle": "2022-04-22T08:19:28.795478Z",
     "shell.execute_reply": "2022-04-22T08:19:28.794551Z",
     "shell.execute_reply.started": "2022-04-22T08:19:28.790818Z"
    }
   },
   "outputs": [],
   "source": [
    "data = []\n",
    "labels = []\n",
    "classes = 43\n",
    "cur_path = os.getcwd()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-22T08:19:28.866185Z",
     "iopub.status.busy": "2022-04-22T08:19:28.865900Z",
     "iopub.status.idle": "2022-04-22T08:19:28.889602Z",
     "shell.execute_reply": "2022-04-22T08:19:28.888702Z",
     "shell.execute_reply.started": "2022-04-22T08:19:28.866157Z"
    }
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ed3c73f8c5f04c74b0e4342ceed3f71c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/43 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in tqdm(range(classes)):\n",
    "    path = os.path.join(cur_path,'Train',str(i))\n",
    "    images = os.listdir(path)\n",
    "\n",
    "    for a in images:\n",
    "        try:\n",
    "            image = Image.open(path + '\\\\'+ a)\n",
    "            image = image.resize((30,30))\n",
    "            image = np.array(image)\n",
    "            data.append(image)\n",
    "            labels.append(i)\n",
    "        except:\n",
    "            print(\"Error loading image\")\n",
    "\n",
    "data = np.array(data)\n",
    "labels = np.array(labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-22T08:19:28.931124Z",
     "iopub.status.busy": "2022-04-22T08:19:28.930846Z",
     "iopub.status.idle": "2022-04-22T08:19:28.946782Z",
     "shell.execute_reply": "2022-04-22T08:19:28.945926Z",
     "shell.execute_reply.started": "2022-04-22T08:19:28.931095Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of data is: (39209, 30, 30, 3)\n",
      "Shape of labels is: (39209,)\n"
     ]
    }
   ],
   "source": [
    "print(\"Shape of data is:\", data.shape)\n",
    "print(\"Shape of labels is:\", labels.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Splitting the data into Train and Test Set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-22T08:19:28.988942Z",
     "iopub.status.busy": "2022-04-22T08:19:28.988411Z",
     "iopub.status.idle": "2022-04-22T08:19:29.010726Z",
     "shell.execute_reply": "2022-04-22T08:19:29.009780Z",
     "shell.execute_reply.started": "2022-04-22T08:19:28.988913Z"
    }
   },
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(data, labels, test_size=0.2, random_state=42)\n",
    "\n",
    "y_train = to_categorical(y_train, 43)\n",
    "y_test = to_categorical(y_test, 43)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-22T08:19:29.107560Z",
     "iopub.status.busy": "2022-04-22T08:19:29.107262Z",
     "iopub.status.idle": "2022-04-22T08:19:29.127210Z",
     "shell.execute_reply": "2022-04-22T08:19:29.126333Z",
     "shell.execute_reply.started": "2022-04-22T08:19:29.107529Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of train images is: (31367, 30, 30, 3)\n",
      "Shape of test images is: (7842, 30, 30, 3)\n",
      "Shape of labels is: (31367, 43)\n",
      "Shape of labels is: (7842, 43)\n"
     ]
    }
   ],
   "source": [
    "print(\"Shape of train images is:\", X_train.shape)\n",
    "print(\"Shape of test images is:\", X_test.shape)\n",
    "print(\"Shape of labels is:\", y_train.shape)\n",
    "print(\"Shape of labels is:\", y_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Building Model CNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-22T08:19:29.136520Z",
     "iopub.status.busy": "2022-04-22T08:19:29.136230Z",
     "iopub.status.idle": "2022-04-22T08:19:29.139991Z",
     "shell.execute_reply": "2022-04-22T08:19:29.139081Z",
     "shell.execute_reply.started": "2022-04-22T08:19:29.136476Z"
    }
   },
   "outputs": [],
   "source": [
    "epochs = 10\n",
    "batch_size = 32"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-22T08:19:29.210878Z",
     "iopub.status.busy": "2022-04-22T08:19:29.210425Z",
     "iopub.status.idle": "2022-04-22T08:19:29.237309Z",
     "shell.execute_reply": "2022-04-22T08:19:29.236377Z",
     "shell.execute_reply.started": "2022-04-22T08:19:29.210849Z"
    }
   },
   "outputs": [],
   "source": [
    "cnn_model = Sequential()\n",
    "cnn_model.add(Conv2D(filters=32, kernel_size=(5,5), activation='relu', input_shape=X_train.shape[1:]))\n",
    "cnn_model.add(Conv2D(filters=32, kernel_size=(5,5), activation='relu'))\n",
    "cnn_model.add(MaxPool2D(pool_size=(2, 2)))\n",
    "cnn_model.add(Dropout(rate=0.25))\n",
    "cnn_model.add(Conv2D(filters=64, kernel_size=(3, 3), activation='relu'))\n",
    "cnn_model.add(Conv2D(filters=64, kernel_size=(3, 3), activation='relu'))\n",
    "cnn_model.add(MaxPool2D(pool_size=(2, 2)))\n",
    "cnn_model.add(Dropout(rate=0.25))\n",
    "cnn_model.add(Flatten())\n",
    "cnn_model.add(Dense(256, activation='relu'))\n",
    "cnn_model.add(Dropout(rate=0.5))\n",
    "cnn_model.add(Dense(43, activation='softmax'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Summary of the CNN Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-22T08:19:29.258219Z",
     "iopub.status.busy": "2022-04-22T08:19:29.257934Z",
     "iopub.status.idle": "2022-04-22T08:19:29.277610Z",
     "shell.execute_reply": "2022-04-22T08:19:29.276478Z",
     "shell.execute_reply.started": "2022-04-22T08:19:29.258193Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d (Conv2D)              (None, 26, 26, 32)        2432      \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 22, 22, 32)        25632     \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 11, 11, 32)        0         \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 11, 11, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 9, 9, 64)          18496     \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 7, 7, 64)          36928     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 3, 3, 64)          0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 3, 3, 64)          0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 576)               0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 256)               147712    \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 43)                11051     \n",
      "=================================================================\n",
      "Total params: 242,251\n",
      "Trainable params: 242,251\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "cnn_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "execution": {
     "iopub.status.busy": "2022-04-22T08:19:29.278463Z",
     "iopub.status.idle": "2022-04-22T08:19:29.279250Z",
     "shell.execute_reply": "2022-04-22T08:19:29.279024Z",
     "shell.execute_reply.started": "2022-04-22T08:19:29.278997Z"
    }
   },
   "outputs": [],
   "source": [
    "cnn_model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train the CNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-22T08:19:29.284198Z",
     "iopub.status.busy": "2022-04-22T08:19:29.284011Z",
     "iopub.status.idle": "2022-04-22T08:19:29.302384Z",
     "shell.execute_reply": "2022-04-22T08:19:29.301367Z",
     "shell.execute_reply.started": "2022-04-22T08:19:29.284175Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "WARNING:tensorflow:AutoGraph could not transform <bound method Dense.call of <keras.layers.core.Dense object at 0x000001AD911C8310>> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: invalid syntax (tmpb6em6_m8.py, line 48)\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "WARNING: AutoGraph could not transform <bound method Dense.call of <keras.layers.core.Dense object at 0x000001AD911C8310>> and will run it as-is.\n",
      "Please report this to the TensorFlow team. When filing the bug, set the verbosity to 10 (on Linux, `export AUTOGRAPH_VERBOSITY=10`) and attach the full output.\n",
      "Cause: invalid syntax (tmpb6em6_m8.py, line 48)\n",
      "To silence this warning, decorate the function with @tf.autograph.experimental.do_not_convert\n",
      "981/981 [==============================] - 71s 54ms/step - loss: 2.8107 - accuracy: 0.3929 - val_loss: 0.2284 - val_accuracy: 0.9380\n",
      "Epoch 2/10\n",
      "981/981 [==============================] - 54s 55ms/step - loss: 0.4719 - accuracy: 0.8764 - val_loss: 0.0966 - val_accuracy: 0.9770\n",
      "Epoch 3/10\n",
      "981/981 [==============================] - 52s 53ms/step - loss: 0.2966 - accuracy: 0.9213 - val_loss: 0.1034 - val_accuracy: 0.9740\n",
      "Epoch 4/10\n",
      "981/981 [==============================] - 52s 53ms/step - loss: 0.2427 - accuracy: 0.9398 - val_loss: 0.0647 - val_accuracy: 0.9853\n",
      "Epoch 5/10\n",
      "981/981 [==============================] - 52s 53ms/step - loss: 0.2550 - accuracy: 0.9352 - val_loss: 0.0768 - val_accuracy: 0.9795\n",
      "Epoch 6/10\n",
      "981/981 [==============================] - 60s 62ms/step - loss: 0.2403 - accuracy: 0.9398 - val_loss: 0.0702 - val_accuracy: 0.9828\n",
      "Epoch 7/10\n",
      "981/981 [==============================] - 55s 56ms/step - loss: 0.2299 - accuracy: 0.9430 - val_loss: 0.0893 - val_accuracy: 0.9768\n",
      "Epoch 8/10\n",
      "981/981 [==============================] - 52s 53ms/step - loss: 0.2218 - accuracy: 0.9441 - val_loss: 0.0552 - val_accuracy: 0.9884\n",
      "Epoch 9/10\n",
      "981/981 [==============================] - 58s 59ms/step - loss: 0.1798 - accuracy: 0.9548 - val_loss: 0.0556 - val_accuracy: 0.9874\n",
      "Epoch 10/10\n",
      "981/981 [==============================] - 63s 64ms/step - loss: 0.2099 - accuracy: 0.9515 - val_loss: 0.0375 - val_accuracy: 0.9904\n"
     ]
    }
   ],
   "source": [
    "cnn_history = cnn_model.fit(X_train, y_train, batch_size=batch_size, epochs=epochs, validation_data=(X_test, y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "execution": {
     "iopub.status.busy": "2022-04-22T08:19:29.303651Z",
     "iopub.status.idle": "2022-04-22T08:19:29.304332Z",
     "shell.execute_reply": "2022-04-22T08:19:29.304106Z",
     "shell.execute_reply.started": "2022-04-22T08:19:29.304076Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'loss': [1.5172663927078247,\n",
       "  0.4031127095222473,\n",
       "  0.28953370451927185,\n",
       "  0.2362450808286667,\n",
       "  0.23423069715499878,\n",
       "  0.22041629254817963,\n",
       "  0.23476025462150574,\n",
       "  0.2176714837551117,\n",
       "  0.1991080790758133,\n",
       "  0.20416465401649475],\n",
       " 'accuracy': [0.6194407939910889,\n",
       "  0.8950489163398743,\n",
       "  0.9246660470962524,\n",
       "  0.9400962591171265,\n",
       "  0.9415627717971802,\n",
       "  0.9444639086723328,\n",
       "  0.9432205557823181,\n",
       "  0.9465361833572388,\n",
       "  0.9516689777374268,\n",
       "  0.9527209997177124],\n",
       " 'val_loss': [0.22844165563583374,\n",
       "  0.09659615904092789,\n",
       "  0.10337609052658081,\n",
       "  0.06471280008554459,\n",
       "  0.07677768170833588,\n",
       "  0.07021987438201904,\n",
       "  0.08931980282068253,\n",
       "  0.055218107998371124,\n",
       "  0.05555613711476326,\n",
       "  0.037519421428442],\n",
       " 'val_accuracy': [0.9380260109901428,\n",
       "  0.9770466685295105,\n",
       "  0.9739862084388733,\n",
       "  0.9853353500366211,\n",
       "  0.9794695377349854,\n",
       "  0.9827849864959717,\n",
       "  0.9767916202545166,\n",
       "  0.9883958101272583,\n",
       "  0.9873756766319275,\n",
       "  0.9904361367225647]}"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cnn_history.history"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Display the Accuracy and Loss Plot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CNN training loss:  0.199\n",
      "CNN Val loss:  0.038\n",
      "CNN accuracy: 95.27 %\n",
      "CNN Val accuracy: 99.044 %\n"
     ]
    }
   ],
   "source": [
    "print(\"CNN training loss: \", round(min(cnn_history.history['loss']),3))\n",
    "print(\"CNN Val loss: \", round(min(cnn_history.history['val_loss']),3))\n",
    "print(\"CNN accuracy:\", round(max(cnn_history.history['accuracy']),4)*100,\"%\")\n",
    "print(\"CNN Val accuracy:\", round(max(cnn_history.history['val_accuracy']),5)*100,\"%\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-22T08:19:29.311704Z",
     "iopub.status.busy": "2022-04-22T08:19:29.311329Z",
     "iopub.status.idle": "2022-04-22T08:19:29.329961Z",
     "shell.execute_reply": "2022-04-22T08:19:29.329000Z",
     "shell.execute_reply.started": "2022-04-22T08:19:29.311676Z"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(cnn_history.history['accuracy'], label='training accuracy')\n",
    "plt.plot(cnn_history.history['val_accuracy'], label='val accuracy')\n",
    "plt.title('Accuracy')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('accuracy')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "execution": {
     "iopub.status.busy": "2022-04-22T08:19:29.330879Z",
     "iopub.status.idle": "2022-04-22T08:19:29.331711Z",
     "shell.execute_reply": "2022-04-22T08:19:29.331486Z",
     "shell.execute_reply.started": "2022-04-22T08:19:29.331462Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(cnn_history.history['loss'], label='training loss')\n",
    "plt.plot(cnn_history.history['val_loss'], label='val loss')\n",
    "plt.title('Loss')\n",
    "plt.xlabel('epochs')\n",
    "plt.ylabel('loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Test model with test dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-22T08:19:29.334958Z",
     "iopub.status.busy": "2022-04-22T08:19:29.334269Z",
     "iopub.status.idle": "2022-04-22T08:19:29.372845Z",
     "shell.execute_reply": "2022-04-22T08:19:29.372220Z",
     "shell.execute_reply.started": "2022-04-22T08:19:29.334922Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "y_test = pd.read_csv('Test.csv')\n",
    "labels = y_test[\"ClassId\"].values\n",
    "imgs = y_test[\"Path\"].values\n",
    "data=[]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2022-04-22T08:19:29.374536Z",
     "iopub.status.busy": "2022-04-22T08:19:29.374281Z",
     "iopub.status.idle": "2022-04-22T08:19:29.394593Z",
     "shell.execute_reply": "2022-04-22T08:19:29.392963Z",
     "shell.execute_reply.started": "2022-04-22T08:19:29.374477Z"
    }
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3ae057e4637647488adcc50c6ec610d6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/12630 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for img in tqdm(imgs):\n",
    "    image = Image.open(img)\n",
    "    image = image.resize((30,30))\n",
    "    data.append(np.array(image))\n",
    "X_test=np.array(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Shape of train images is: (31367, 30, 30, 3)\n",
      "Shape of test images is: (12630, 30, 30, 3)\n",
      "Shape of labels is: (31367, 43)\n",
      "Shape of labels is: (12630, 8)\n"
     ]
    }
   ],
   "source": [
    "print(\"Shape of train images is:\", X_train.shape)\n",
    "print(\"Shape of test images is:\", X_test.shape)\n",
    "print(\"Shape of labels is:\", y_train.shape)\n",
    "print(\"Shape of labels is:\", y_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Make Prediction"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "execution": {
     "iopub.status.busy": "2022-04-22T08:19:29.395346Z",
     "iopub.status.idle": "2022-04-22T08:19:29.397115Z",
     "shell.execute_reply": "2022-04-22T08:19:29.396938Z",
     "shell.execute_reply.started": "2022-04-22T08:19:29.396915Z"
    }
   },
   "outputs": [],
   "source": [
    "cnn_pred = cnn_model.predict_classes(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy:  96.785 %\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "accuracy_score = accuracy_score(labels, cnn_pred)\n",
    "print(\"Accuracy: \", round(accuracy_score,5)*100, \"%\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Save CNN Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "cnn_model.save(\"traffic_sign_classification.h5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}