sveska_2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "0fdb8420",
   "metadata": {},
   "source": [
    "# PREPOZNAVANJE SAOBRACAJNIH ZNAKOVA"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d1973d0a",
   "metadata": {},
   "source": [
    "Skup podataka o slikama sastoji se od vise od 30 000 slika razlicitih saobracajnih znakova. Prisutne su 43 razlicite klase  u skupu podataka za klasifikaciju slika."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "982e4828",
   "metadata": {},
   "source": [
    "- **0** 'Ogranicenje brzine (20km/h)  **1** 'Ogranicenje brzine (30km/h)'\n",
    "- **2** 'Ogranicenje brzine (50km/h)' **3** 'Ogranicenje brzine (60km/h)'\n",
    "- **4** 'Ogranicenje brzine (70km/h)') **5** 'Ogranicenje brzine (80km/h)'\n",
    "- **6** 'Prestanak ogranicenja brzine (80km/h)' **7** 'Ogranicenje brzine (100km/h)'\n",
    "- **8** 'Ogranicenje brzine (120km/h)' **9** Zabrana preticanja za motorna vozila'\n",
    "- **10** 'Zabrana preticanja za teretna vozila'\n",
    "- **11** 'Spajanje sa sporednim putem pod pravim uglom sa desne strane' **12** 'Put sa prvenstvom prolaza'\n",
    "- **13** 'Ukrstanje sa putem sa prvenstvom prolaza' **14** 'Stop' **15** 'Zabrana saobracaja u oba smera'\n",
    "- **16** 'Zabrana saobracaja za teretna vozila') **17** 'Zabrana saobracaja u jednom smeru'\n",
    "- **18** 'Opasnost na putu') **19** 'Krivina nalevo' **20** 'Krivina nadesno' \n",
    "- **21** 'Dvostruka krivina' **22** 'Neravan kolovoz' **23** 'Klizav kolovoz'\n",
    "- **24** Suzenje puta sa desne strane' **25** 'Radovi na putu'\n",
    "- **26** Nailazenje na semafor' **27** 'Pesaci na putu' **28** 'Deca na putu'\n",
    "- **29** Biciklisti na putu' **30** 'Poledica/sneg' **31** 'Divljac na putu'\n",
    "- **32** 'Prestanak svih zabrana' **33** 'Obavezan smer - poludesno'\n",
    "- **34** 'Obavezan smer - polulevo' **35** 'Obazvezan smer - pravo' **36** 'Dozvoljeni smerovi (pravo ili desno)'\n",
    "- **37** Dozvoljeni smerovi (pravo ili levo)' \n",
    "- **38** 'Obavezno obilazenje s desne strane' **39** 'Obavezno obilazenje s leve strane' **40** Kruzni tok saobracaja' \n",
    "- **41** 'Prestanak zabrane preticanja svih motornih vozila, osim motocikala bez prikolice'\n",
    "- **42** Prestanak zabrane preticanja za teretna vozila'"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1d793525",
   "metadata": {},
   "source": [
    "Ucitavanje biblioteka."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "0c90c2de",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "import pandas as pd \n",
    "import numpy as np \n",
    "import matplotlib.pyplot as plt \n",
    "import random"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "45cc7b63",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.utils import shuffle\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import confusion_matrix, classification_report"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "01a363cf",
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Conv2D, MaxPooling2D, Dense, Flatten, Dropout, Input\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "from tensorflow.keras.losses import CategoricalCrossentropy\n",
    "from tensorflow.keras.models import load_model\n",
    "from tensorflow.keras import utils"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3256cde2",
   "metadata": {},
   "source": [
    "Ucitavanje skupa podataka."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "f224d659",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"./traffic-signs-data/train.p\", mode='rb') as training_data:\n",
    "    train = pickle.load(training_data)\n",
    "with open(\"./traffic-signs-data/valid.p\", mode='rb') as validation_data:\n",
    "    valid = pickle.load(validation_data)\n",
    "with open(\"./traffic-signs-data/test.p\", mode='rb') as testing_data:\n",
    "    test = pickle.load(testing_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "c200f912",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, y_train = train['features'], train['labels']\n",
    "X_validation, y_validation = valid['features'], valid['labels']\n",
    "X_test, y_test = test['features'], test['labels']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "46594fdb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(34799, 32, 32, 3)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "58b6037b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(34799,)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "f5462797",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.title('Saobracajni znakovi u test skupu')\n",
    "plt.barh(np.arange(0, 43), np.bincount(y_test), color='purple')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ed908dc0",
   "metadata": {},
   "source": [
    "Jedna instanca ucitanog skupa."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "ad7be604",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "36"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD5CAYAAADhukOtAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAAAasUlEQVR4nO2dW4wkZ3XH/6equue69nrxhZWxYkB+CELBoJGF5AiRkCAHIRkeQPgB+cFiecBSkMiD5UjBeSNRAPGEtMQWJiKAFUBYkZVgWYmsSJHD4hhfsoSbHHC88vqy9s7MznTX5eShy9LafP8zsz3TPRt//5+02p46/VWd+rpOV/f373OOuTuEEG98ioN2QAgxHxTsQmSCgl2ITFCwC5EJCnYhMkHBLkQmVHsZbGY3AfgKgBLA37r7F6LnF0XhZVmm9xWMY/JgLBryPbrx97ii4lNSVYP09iJ9TgBQFMGZectN3gTjuImed3DOFtmCI3nH/e/atK1pajqmCfYXEszHVNfOtHMVXgf8urIyPculhSeW3DwabaOp6+QObVqd3cxKAD8D8McAngHwIwC3uPt/sTGDwcCPXPampK0KPmN0TfrC3ybbJztc5Purlqht6bLLqe2Ky69Kbr/skkvomNWFIbVZu0Ft9ehFanMSSADgSL8hWcnPeTg8RG1lcHnUW69Q27mzLyW3P//Cc3TMS+tnqa2L3nYCH9s6/eZSB2Os5NdOOVymtuHSYWpbXD3CbYfSbwSHhh0dY/U4uf2pxx/F5sZ6crL28jH+BgC/cPdfufsYwLcB3LyH/QkhZshegv1qAL857+9n+m1CiIuQvXxnT31U+K0PR2Z2DMAxACgKrQcKcVDsJfqeAXDNeX+/BcCzr3+Sux939zV3X1OwC3Fw7CX6fgTgOjN7q5kNAXwCwP3745YQYr+Z+mO8uzdmdjuAf8ZEervH3Z/aeWB6hXFMVk0BoOsuXD5p621qKwJJY7SeXkUGgA0iyw2JnAgAhfNV5OWFFWpbWOKr+JHkZUj7Upb8pWZqBwDUI75CvrlxmtpePJO2nd3c5McKVIY2mMciWlknUll0l7Ng5T+S0KI5tkBMGBJ5NnqdWRxF7Elnd/cHADywl30IIeaDvkQLkQkKdiEyQcEuRCYo2IXIBAW7EJmwp9X4C8VgqIp0osa5cfqH/QDQIS0zWKBnlIGtC2Q+Ny7ZbbySluWajssg40t5AkS9wpMqVpd44sogkPpAfKnH5+iQdszlsNEml95eeXmd2s6up1/PUZC71ARqUpRtVoTzkZavSKIZAMCDayf6YVgkBVuQcDYepedquBhkbtL98ePozi5EJijYhcgEBbsQmaBgFyITFOxCZMJcV+NhBpTpBI+i4qvxK8sLye3tmK+cN8Gyb00Sayb7HFEbWDJGsNK6HtiiFVp3/tKsLKbnAwAGJMnHa35e2+d4CaxXXj5DbZtbfI7d06qLR3X3AlUD0Sp4kCTD6sl1wbEsOFaUJBPZmuC1XhqmlZcWXDVix/IoiYdahBBvKBTsQmSCgl2ITFCwC5EJCnYhMkHBLkQmzFV669yxTTpZlAOe+OGWTnS45BAf05U88ePMBu/E4oEshyYtX3mg1m0GtcLqLmqFxG31mNenW0krXsB2kLRyhie7bGzxk6vbSDpMy3IWSG9A1OmG35cCN2AkicqjhJGgRiHbHwB4x6XINqgnNyayXBUk/wyqtPwa1s+jFiHEGwoFuxCZoGAXIhMU7EJkgoJdiExQsAuRCXuS3szsaQDrmGgmjbuvRc/3rkPDMtUGvBWSESlkPWjxVAV1xJy02wFAa7hNBqblsLaO5BguoY0b7r+1QU2+5UVqW2/T+2y2t+iYZhxJRkGGYNA2qqMyFN9fGbwuoVRGLZw2ep0DCa0IbFFLpthHdt58VFWl5eioLuN+6Ox/4O4v7MN+hBAzRB/jhciEvQa7A/ihmf3YzI7th0NCiNmw14/xN7r7s2Z2JYAHzeyn7v7w+U/o3wSOTR7rg4QQB8Weos/dn+3/Pw3g+wBuSDznuLuvuftatHgghJgtUwe7ma2Y2aFXHwP4IIAn98sxIcT+speP8VcB+H5/t64A/L27/1M0wAAYkVDahmepdV36E0HUiqcO3sZi6S0wtcQYZEmx9kOTYdyP7fVgPkZ8n4NBOu2trC7lY5Z5+6SF6H4QynJp6XC8zVtNjetAHgykyEje7IiPUaumoDZkmL1mkWQXHJG2coq+9lLbDKQ3d/8VgHdNO14IMV+0YiZEJijYhcgEBbsQmaBgFyITFOxCZMJcC05aYRgO09JQR7K1AKAlakfUDy2qGxlpK2H/NfKjoEjKK0r+flpGtopntg0WL6E2I8U5t8dcMmLzCyCUIovgXlGW6SxGW+B96qohvwaKjhfM7Ea8gCjL6AsS1MIMu3jgdDYmRxfktZzYyNwHqrLu7EJkgoJdiExQsAuRCQp2ITJBwS5EJsx1NR7gi4VVwV0pSPJBHdRAa4O3sajWWZSEy/JgisEyHTMMVs4HJW/jVFbcVjd8tXhUp9s1NcFcTY3zSSbdn+DRinWQULQ4XKW2lSWuXDT2SnL71ihINAougqi9UphewyYEgJPrO7pOQ5mEoDu7EJmgYBciExTsQmSCgl2ITFCwC5EJCnYhMmG+0puDtlfqIkmjTCfPWCA/FB5kd0Ql46IaXoO0HFYND9ExVhzmBzM+/dsNr6u2Req7AcDiQlqGuvLIETqmDOZxtMlrxm1vcx9HbXqSa+fJHd5x22gcyKWk7h4AVMO0j6XzpBuEKmWoy1G6ls+VsWu15Y5UA9L+ibugO7sQuaBgFyITFOxCZIKCXYhMULALkQkKdiEyYUfpzczuAfBhAKfd/Z39tiMAvgPgWgBPA/i4u5/ZcV9wlET3itQOloVUVUHWFZF+AKCLatCVPIOqIplX5XCJjonqzEWZaFs1l9ewwKWmxUvStd+Wlnntt0OD4DJY4uNeeimdUQYAL26kWzlF5e7aQDhqgqKCG2NuGw7Sr+fywhV0TGV87keBH7SNE2LprR6lZUAn9RoBYFSnx3RE2gZ2d2f/OoCbXrftDgAPuft1AB7q/xZCXMTsGOx9v/WXXrf5ZgD39o/vBfCR/XVLCLHfTPud/Sp3PwUA/f9X7p9LQohZMPOfy5rZMQDHAKAMWiwLIWbLtNH3nJkdBYD+/9Psie5+3N3X3H2tIE0WhBCzZ9pgvx/Arf3jWwH8YH/cEULMit1Ib98C8H4Al5vZMwA+D+ALAO4zs9sA/BrAx3Z9RHJzN9bOJhgTtdQpS55B1QXySWFcehsO0tltwyEvODkO5LVxIK8tDLnktXQoLa8BwPIw/ZIGnYRQB1lvVcUHLiwHLapG6XOrg15TbSSJBp8K3QPJbpyWr2qSwQgAw2Dui5oXqiyDD65tUHByezu9zzPBtbNMst4i6W3HYHf3W4jpAzuNFUJcPGjFTIhMULALkQkKdiEyQcEuRCYo2IXIhLn3emNEfdu8SrtZkkKUAHhjNsQyn1V8n8OFtFxjwbGacSA1Be262prPR3eWyz81KQK5RSQ5AFgc8vmookyuOsg2I3Jk3aSz4QCgGfPMsKBGaHzHIv6PmuA1c+7HMNDXPJB0x8HxlpfS19zictBDkJx0JFHqzi5EJijYhcgEBbsQmaBgFyITFOxCZIKCXYhMmKv05gA6IqIUJXelJVlNbRNlr/H9eSC9VYH0xmgbLq91bZTDH/U9izK5uIzDZmS4xDPUukBqOre+QW31Nj/vETnvOshU7KLMNmqJYTJrtD8r+OtixiVRC3YaXY9lkbYNgmtxYZA+ryIoEKM7uxCZoGAXIhMU7EJkgoJdiExQsAuRCfNNhDEDyMoj7fEEoCzSNu+CVfCglZCT/U3cCNr7kDXcaVeKo/faqE5eWKSXZNdskZpwADBY5jtcDGrQdc732REFJUo0WiKJRgCAoIbbaIsnBrmn5zjIQYpr4QW1DYtgp0VwRCfnVlbBddqlFZSoBZXu7EJkgoJdiExQsAuRCQp2ITJBwS5EJijYhciE3bR/ugfAhwGcdvd39tvuAvApAM/3T7vT3R/YaV/uQE1qcVnBZZeKSGVNJL21QU274C0ubD5JTJHcEfVdihInYi68FVIZSF4eOLLtfI7rYB7pHgPZc3GRJ+ssDrmPoyKQ5c6lJaoglyisJde2UfJVpIlGYh+xBXNvxGaBELybO/vXAdyU2P5ld7++/7djoAshDpYdg93dHwbw0hx8EULMkL18Z7/dzB43s3vM7LJ980gIMROmDfavAng7gOsBnALwRfZEMztmZifM7IQHhQuEELNlqmB39+fcvfVJ9H4NwA3Bc4+7+5q7r4U92IUQM2Wq6DOzo+f9+VEAT+6PO0KIWbEb6e1bAN4P4HIzewbA5wG838yux0RheRrAp3d3OKeZY4ag3RFRICKZoQpqcdWBVBa1oVog8k+g1MAD+SR6r41EnMjGXInOqwnaPw2XV/jBgvZbRibFglqDbjyLriy5NHvF4cPUton15PYXNzfpmCa4FkuLsteoCdFrzeRSC2oUMmk58mHHYHf3WxKb795pnBDi4kJfooXIBAW7EJmgYBciExTsQmSCgl2ITJhvwUnwYoms6B7Ai0dWQZueIngfa0lRRgDogkw69gPASAKMJbTpxoUlLonJKi5dLS5x22UlP9bgUu7liDgyRpBFF7wuUdKYkUxKAFhZWUpuPzvmRSqbhrfD6gIptQjSKT0ItY7oZa+scx9XFtLXfqT+6c4uRCYo2IXIBAW7EJmgYBciExTsQmSCgl2ITJi79MbScjwQm1gefBFIb6znWX+w6WzEWAb9vyzww7uomEdUVJKPYhl4baBdFSW3WTOitvVAGjpLFKpzgTzVBpmKhy/l2XfDjsu22xvp7La6DiTWoEholMUYZj+GWmo6e3BhmReAGhK51FgvRejOLkQ2KNiFyAQFuxCZoGAXIhMU7EJkwtxX4zu2Am1BeyKylDkOVlSLKau4RePYCm5V8Wm0Mng/DVbjw9ZQoS3tf9Siqq63qW3j3Flq29rgK/UbTXpOtp0n3SytXkJtzTZ/rV84e4aP20r7WASr1gPwWnhtw/0og9ZWLNkFAIyEoVs6iQcAllbTK/VFUONPd3YhMkHBLkQmKNiFyAQFuxCZoGAXIhMU7EJkwm7aP10D4BsA3oxJJbDj7v4VMzsC4DsArsWkBdTH3Z1rIDvQNFzuMJIgUQaJE13URDKwmXM5qa7T8kkLnqTRYWEaN2i9OwCwUMYhiUaBXFfzkmsoA9uYv2Ro2rREVZRcumpHvCXT+ogfbLzFx5VkPoqgFh7aLWrqOn59tJHKWkayXPrc3AJJdCsdam1QQ3E3d/YGwOfc/XcBvBfAZ8zsHQDuAPCQu18H4KH+byHERcqOwe7up9z90f7xOoCTAK4GcDOAe/un3QvgIzPyUQixD1zQd3YzuxbAuwE8AuAqdz8FTN4QAFy5794JIfaNXf9c1sxWAXwXwGfd/ayxAvC/Pe4YgGP942l8FELsA7u6s5vZAJNA/6a7f6/f/JyZHe3tRwGcTo119+Puvubua0WwoCaEmC07Rp9Nbsd3Azjp7l86z3Q/gFv7x7cC+MH+uyeE2C928zH+RgCfBPCEmT3Wb7sTwBcA3GdmtwH4NYCP7eqI5JN8EWQMOdGhmqDtT0QZZAaVQbqZEbnGO/6eWViQXRXUOuuCrzyhxEOGDaIiaOOgJdMoKqwW1N5jtfBaruU157isFWmRZSSlkgw277i8hjbIemt5vTsLahGW1SIfR1qfVVH23TaR5YJ6fDsGu7v/G3hO6Ad2Gi+EuDjQl2ghMkHBLkQmKNiFyAQFuxCZoGAXIhPmXnCStnIKKyySIopT/yCPH6sJCgp2DfG94hLJwpBnZA2MyzFtybPlOqavASgsLW112+t0TNS+ahhka9WBHEbbeQXpd0XQogqBDAXncp6TLMaiC8YEmWNR4c7ociyDwqNsXBXIrxUxRT9S1Z1diExQsAuRCQp2ITJBwS5EJijYhcgEBbsQmTBX6c1gqIq0zDMOsomY2MFkPCDOorPAVhS8F5kj3Y8uSKIDCp5d1QQySdfx3ndV4H/J5KuGZ5SNuTqIpuNS09ije0XaVgWyZ9QWL8pGhAdyKZHRPMhso/0IgbjPXiC+2RT9BVnRTgCoqCQaHUcIkQUKdiEyQcEuRCYo2IXIBAW7EJkw50QYB9gKY9SfiFSljRdGg1XJYNk3yAlhboTvmI3x1f1qhbeNqkeBOjHmbYEWyQlEZbzrmh9rHKxMe6RqkO3RqvogsLVjrmrUgdLgSJ+bRf21ogSfSBQIjNG12hLFo22DhC2SDBWGUeCDEOINhIJdiExQsAuRCQp2ITJBwS5EJijYhciEHaU3M7sGwDcAvBlAB+C4u3/FzO4C8CkAz/dPvdPdH4j21bljRBJerOCuMNkokhnKimtogWIUJlWwA3bBNK6sHKG21UsvpbbNjTPc1nAZqiWqURFIgAgSWgZR0k3JJbu6Pke28wSU8ZjXhYuSUwqSXAUANFcq1G1nQXDRsWKKQZFFm6IA42509gbA59z9UTM7BODHZvZgb/uyu//NBR9VCDF3dtPr7RSAU/3jdTM7CeDqWTsmhNhfLug7u5ldC+DdAB7pN91uZo+b2T1mdtl+OyeE2D92HexmtgrguwA+6+5nAXwVwNsBXI/Jnf+LZNwxMzthZieinxMKIWbLroLdzAaYBPo33f17AODuz7l765Pm6V8DcENqrLsfd/c1d1+Lfp8thJgtOwa7TSL0bgAn3f1L520/et7TPgrgyf13TwixX+xmNf5GAJ8E8ISZPdZvuxPALWZ2PSYixtMAPr3TjswMVqUPWQT15Nin/+hLQcM0KMTSWxl8+mCmIkiVG414Rlbz4vPUVlX87A6tLlObWdqXKMlr61xaJgOArU1eoG47yDZru7QsN21mWHRfYlljAFCQFy2q42fBXMWfTaerQUctUasscl6Rf7tZjf83so9QUxdCXFzoF3RCZIKCXYhMULALkQkKdiEyQcEuRCbMueAkp2DVHAF0REaL3qk8kN4CZQUeCEDMDy/4GDOeRReojRgMw9Q8aimrheT2paVFOmblEM+IGxjPUtvaCGzE1Ia/ogzOOTBFsi2TqKKrIGoPVoXZZtNlqTFZLvoRmrF4ieaJm4QQbyQU7EJkgoJdiExQsAuRCQp2ITJBwS5EJsxVenMYnLy/lEHByYqkITVN0A8t8CMqUBgNZMlV3gXTWKSlMAAw5wUW6zHPRLOS6ysdk2sa/r5+eIVn0R0q+bm90nBZkfWPc5KVByBMRwwFu2luWaFcFwwLxkWSbiQtO5HluqAQKGsDF173gU0I8QZCwS5EJijYhcgEBbsQmaBgFyITFOxCZMJcpTcDz1DqgqygQTVIGzou/RQ7lAak4wJZru3S+xwsHqJjynKV2i5ZDWSc7jS1rZ/lfeBGm+kikNub6/xY55aobSGQmra2ecHJjs7/dFlvURVysyjrkPUJnC57rQ2uuShjsmm4MMcyHFumrwHg92nuu+7sQmSCgl2ITFCwC5EJCnYhMkHBLkQm7Lgab2aLAB4GsNA//x/c/fNmdgTAdwBci0n7p4+7O18mfnV/ZHsX9Cdq2I/+oxXaoNZZ3GAyak+U9rHu+Kr0cJHXfotWdocVVwUOr/LV/81z6eSa0Zgn3Yy3eIunUfC6OFEnAL4yXQRZJgVpDRbtb0cb7R0W1Brs+DkHpQ3jWoQtv0asTdcAbNptOmZcp4/lweu1mzv7CMAfuvu7MGnPfJOZvRfAHQAecvfrADzU/y2EuEjZMdh9wkb/56D/5wBuBnBvv/1eAB+ZhYNCiP1ht/3Zy76D62kAD7r7IwCucvdTAND/f+XMvBRC7JldBbu7t+5+PYC3ALjBzN652wOY2TEzO2FmJ6LvE0KI2XJBq/Hu/jKAfwVwE4DnzOwoAPT/J3/f6e7H3X3N3deiKh9CiNmyY/SZ2RVmdrh/vATgjwD8FMD9AG7tn3YrgB/MyEchxD6wm0SYowDuNbMSkzeH+9z9H83s3wHcZ2a3Afg1gI/txREP5A6W5xCUYgvTLUI/gkpigyr93ujgbZCa8cvUNg6c9JZLZWgDyW6QlnEWlnmduS6QhertLWobjXgNwJoU7GN1/ICdagpO94oySTcSX7tAUkRQQy9UdINkHe/Sr3XbcOmtYzUFA0lxx2B398cBvDux/UUAH9hpvBDi4kBfooXIBAW7EJmgYBciExTsQmSCgl2ITDCaFTSLg5k9D+B/+j8vB/DC3A7OkR+vRX68lv9vfvyOu1+RMsw12F9zYLMT7r52IAeXH/IjQz/0MV6ITFCwC5EJBxnsxw/w2OcjP16L/Hgtbxg/Duw7uxBivuhjvBCZcCDBbmY3mdl/m9kvzOzAateZ2dNm9oSZPWZmJ+Z43HvM7LSZPXnetiNm9qCZ/bz//7ID8uMuM/vffk4eM7MPzcGPa8zsX8zspJk9ZWZ/2m+f65wEfsx1Tsxs0cz+w8x+0vvxl/32vc2Hu8/1H4ASwC8BvA3AEMBPALxj3n70vjwN4PIDOO77ALwHwJPnbftrAHf0j+8A8FcH5MddAP5szvNxFMB7+seHAPwMwDvmPSeBH3OdE0wycFf7xwMAjwB4717n4yDu7DcA+IW7/8rdxwC+jUnxymxw94cBvPS6zXMv4En8mDvufsrdH+0frwM4CeBqzHlOAj/mik/Y9yKvBxHsVwP4zXl/P4MDmNAeB/BDM/uxmR07IB9e5WIq4Hm7mT3ef8yf+deJ8zGzazGpn3CgRU1f5wcw5zmZRZHXgwj2VD2Pg5IEbnT39wD4EwCfMbP3HZAfFxNfBfB2THoEnALwxXkd2MxWAXwXwGfd/ey8jrsLP+Y+J76HIq+Mgwj2ZwBcc97fbwHw7AH4AXd/tv//NIDvY/IV46DYVQHPWePuz/UXWgfga5jTnJjZAJMA+6a7f6/fPPc5SflxUHPSH/tlXGCRV8ZBBPuPAFxnZm81syGAT2BSvHKumNmKmR169TGADwJ4Mh41Uy6KAp6vXkw9H8Uc5sQm/bjuBnDS3b90nmmuc8L8mPeczKzI67xWGF+32vghTFY6fwngzw/Ih7dhogT8BMBT8/QDwLcw+ThYY/JJ5zYAb8KkjdbP+/+PHJAffwfgCQCP9xfX0Tn48fuYfJV7HMBj/b8PzXtOAj/mOicAfg/Af/bHexLAX/Tb9zQf+gWdEJmgX9AJkQkKdiEyQcEuRCYo2IXIBAW7EJmgYBciExTsQmSCgl2ITPg/xCXhy0bLa78AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "i = 1001\n",
    "plt.imshow(X_train[i]) \n",
    "y_train[i]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "3d722e05",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(34799, 32, 32, 3)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c3da834f",
   "metadata": {},
   "source": [
    "## PRIPREMA PODATAKA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "240924cf",
   "metadata": {},
   "outputs": [],
   "source": [
    "# mesamo skup podataka\n",
    "X_train, y_train = shuffle(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "ea2bda43",
   "metadata": {},
   "outputs": [],
   "source": [
    "# prebacujemo slike u monohromatske\n",
    "X_train_gray = np.sum(X_train/3, axis=3, keepdims=True)\n",
    "X_test_gray  = np.sum(X_test/3, axis=3, keepdims=True)\n",
    "X_validation_gray  = np.sum(X_validation/3, axis=3, keepdims=True) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "2ea18425",
   "metadata": {},
   "outputs": [],
   "source": [
    "# a potom ih i normalizujemo\n",
    "X_train_gray_norm = (X_train_gray - 128)/128 \n",
    "X_test_gray_norm = (X_test_gray - 128)/128\n",
    "X_validation_gray_norm = (X_validation_gray - 128)/128"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "b32f8f9b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(34799, 32, 32, 1)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_gray.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "ac57146d",
   "metadata": {},
   "outputs": [],
   "source": [
    "# primenicemo one-hot encoding na ciljnu promenljivu\n",
    "y_train = utils.to_categorical(y_train, 43)\n",
    "y_test = utils.to_categorical(y_test, 43)\n",
    "y_validation = utils.to_categorical(y_validation, 43)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "d8434e27",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "i = 610\n",
    "plt.imshow(X_train_gray[i].squeeze(), cmap='gray')\n",
    "plt.figure()\n",
    "plt.imshow(X_train[i])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "6038a875",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(32, 32, 1)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "image_shape = X_train_gray[i].shape\n",
    "image_shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cd35b675",
   "metadata": {},
   "source": [
    "## TRENIRANJE MODELA"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3b0eca6c",
   "metadata": {},
   "source": [
    "Nasa mreza imace sledecu arhitekturu."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "7fd0a417",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(Input(shape=image_shape))\n",
    "model.add(Conv2D(filters=32, kernel_size=(5,5), activation='relu'))\n",
    "model.add(Conv2D(filters=32, kernel_size=(5,5), activation='relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
    "model.add(Dropout(rate=0.25))\n",
    "model.add(Conv2D(filters=64, kernel_size=(3, 3), activation='relu'))\n",
    "model.add(Conv2D(filters=64, kernel_size=(3, 3), activation='relu'))\n",
    "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
    "model.add(Dropout(rate=0.25))\n",
    "model.add(Flatten())\n",
    "model.add(Dense(256, activation='relu'))\n",
    "model.add(Dropout(rate=0.5))\n",
    "model.add(Dense(43, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "edd4ad12",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d (Conv2D)              (None, 28, 28, 32)        832       \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 24, 24, 32)        25632     \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 12, 12, 32)        0         \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 12, 12, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 10, 10, 64)        18496     \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 8, 8, 64)          36928     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 4, 4, 64)          0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 4, 4, 64)          0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 1024)              0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 256)               262400    \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 43)                11051     \n",
      "=================================================================\n",
      "Total params: 355,339\n",
      "Trainable params: 355,339\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f81891cc",
   "metadata": {},
   "source": [
    "Funkcija gubitka koju mreza minimizuje ce biti kategoricka unakrsna entropija. Za treniranje mreze cemo koristiti optimizator Adam sa podesenim korakom ucenja."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "f41625e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(loss=CategoricalCrossentropy(), optimizer=Adam(lr=0.001), metrics =['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8cffc479",
   "metadata": {},
   "source": [
    "Mrezu cemo trenirati u 15 epoha koriscenjem paketica velicine 32 instanci."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "23b6b76b",
   "metadata": {},
   "outputs": [],
   "source": [
    "epochs = 15\n",
    "batch_size = 32"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "1691f57b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/15\n",
      "1088/1088 [==============================] - 1198s 1s/step - loss: 0.9193 - accuracy: 0.7457 - val_loss: 0.1811 - val_accuracy: 0.9469\n",
      "Epoch 2/15\n",
      "1088/1088 [==============================] - 1144s 1s/step - loss: 0.1715 - accuracy: 0.9484 - val_loss: 0.0860 - val_accuracy: 0.9744\n",
      "Epoch 3/15\n",
      "1088/1088 [==============================] - 1137s 1s/step - loss: 0.1208 - accuracy: 0.9628 - val_loss: 0.0678 - val_accuracy: 0.9830\n",
      "Epoch 4/15\n",
      "1088/1088 [==============================] - 1142s 1s/step - loss: 0.0881 - accuracy: 0.9736 - val_loss: 0.0748 - val_accuracy: 0.9814\n",
      "Epoch 5/15\n",
      "1088/1088 [==============================] - 1138s 1s/step - loss: 0.0800 - accuracy: 0.9757 - val_loss: 0.0800 - val_accuracy: 0.9798\n",
      "Epoch 6/15\n",
      "1088/1088 [==============================] - 1136s 1s/step - loss: 0.0758 - accuracy: 0.9778 - val_loss: 0.0796 - val_accuracy: 0.9823\n",
      "Epoch 7/15\n",
      "1088/1088 [==============================] - 1136s 1s/step - loss: 0.0670 - accuracy: 0.9799 - val_loss: 0.0603 - val_accuracy: 0.9850\n",
      "Epoch 8/15\n",
      "1088/1088 [==============================] - 1137s 1s/step - loss: 0.0664 - accuracy: 0.9805 - val_loss: 0.0564 - val_accuracy: 0.9878\n",
      "Epoch 9/15\n",
      "1088/1088 [==============================] - 1139s 1s/step - loss: 0.0525 - accuracy: 0.9845 - val_loss: 0.0559 - val_accuracy: 0.9896\n",
      "Epoch 10/15\n",
      "1088/1088 [==============================] - 1135s 1s/step - loss: 0.0634 - accuracy: 0.9820 - val_loss: 0.0591 - val_accuracy: 0.9889\n",
      "Epoch 11/15\n",
      "1088/1088 [==============================] - 1135s 1s/step - loss: 0.0493 - accuracy: 0.9857 - val_loss: 0.0626 - val_accuracy: 0.9878\n",
      "Epoch 12/15\n",
      "1088/1088 [==============================] - 1135s 1s/step - loss: 0.0495 - accuracy: 0.9856 - val_loss: 0.1065 - val_accuracy: 0.9812\n",
      "Epoch 13/15\n",
      "1088/1088 [==============================] - 1137s 1s/step - loss: 0.0468 - accuracy: 0.9860 - val_loss: 0.0627 - val_accuracy: 0.9896\n",
      "Epoch 14/15\n",
      "1088/1088 [==============================] - 1144s 1s/step - loss: 0.0494 - accuracy: 0.9862 - val_loss: 0.0624 - val_accuracy: 0.9864\n",
      "Epoch 15/15\n",
      "1088/1088 [==============================] - 1140s 1s/step - loss: 0.0444 - accuracy: 0.9874 - val_loss: 0.0518 - val_accuracy: 0.9896\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(X_train_gray_norm, y_train, batch_size=batch_size, epochs=epochs, verbose=1, \n",
    "                    validation_data = (X_validation_gray_norm, y_validation))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c58a42d0",
   "metadata": {},
   "source": [
    "## EVALUACIJA MODELA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "1bef917e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Accuracy : 0.9718\n"
     ]
    }
   ],
   "source": [
    "score = model.evaluate(X_test_gray_norm, y_test,verbose=0)\n",
    "print('Test Accuracy : {:.4f}'.format(score[1]))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ddbaf6e0",
   "metadata": {},
   "source": [
    "Mozemo i nacrtati grafike funkcije gubitka i tacnosti u toku treniranja."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "9889385f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10, 4))\n",
    "\n",
    "plt.subplot(1, 2, 1)\n",
    "plt.title('Loss')\n",
    "plt.plot(np.arange(0, epochs), history.history['loss'], label='train')\n",
    "plt.plot(np.arange(0, epochs), history.history['val_loss'], label='val')\n",
    "plt.legend(loc='best')\n",
    "\n",
    "plt.subplot(1, 2, 2)\n",
    "plt.title('Accuracy')\n",
    "plt.plot(np.arange(0, epochs), history.history['accuracy'], label='train')\n",
    "plt.plot(np.arange(0, epochs), history.history['val_accuracy'], label='val')\n",
    "plt.legend(loc='best')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d0d8e670",
   "metadata": {},
   "source": [
    "Funkcijom save cemo sacuvati nas model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "62cc2dd9",
   "metadata": {},
   "outputs": [],
   "source": [
    "model_path = 'cnn_model.h5'\n",
    "model.save(model_path)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b785c7c6",
   "metadata": {},
   "source": [
    "Mozemo jos testirati kako se nas model ponasa za novi znak."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "096ab404",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_predicted = model.predict(X_test_gray_norm)\n",
    "y_predicted_classes = np.argmax(y_predicted, axis=1)\n",
    "\n",
    "y_test_classes = np.argmax(y_test, axis = 1) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "0e422a40",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 60   0   0 ...   0   0   0]\n",
      " [  0 715   3 ...   1   0   0]\n",
      " [  0   1 748 ...   1   0   0]\n",
      " ...\n",
      " [  0   0   0 ...  81   0   0]\n",
      " [  0   0   0 ...   0  48   0]\n",
      " [  0   0   0 ...   0   1  89]]\n"
     ]
    }
   ],
   "source": [
    "cm = confusion_matrix(y_test_classes, y_predicted_classes)\n",
    "print(cm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "4bf3695b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       1.00      1.00      1.00        60\n",
      "           1       0.93      0.99      0.96       720\n",
      "           2       0.96      1.00      0.98       750\n",
      "           3       1.00      0.96      0.98       450\n",
      "           4       0.99      0.98      0.98       660\n",
      "           5       0.97      0.94      0.95       630\n",
      "           6       1.00      0.88      0.94       150\n",
      "           7       0.98      0.98      0.98       450\n",
      "           8       0.98      0.99      0.98       450\n",
      "           9       0.97      1.00      0.99       480\n",
      "          10       1.00      1.00      1.00       660\n",
      "          11       0.97      0.97      0.97       420\n",
      "          12       1.00      0.98      0.99       690\n",
      "          13       0.99      1.00      0.99       720\n",
      "          14       1.00      0.98      0.99       270\n",
      "          15       1.00      1.00      1.00       210\n",
      "          16       0.99      0.99      0.99       150\n",
      "          17       0.99      0.99      0.99       360\n",
      "          18       0.99      0.91      0.95       390\n",
      "          19       0.83      0.88      0.85        60\n",
      "          20       0.90      0.99      0.94        90\n",
      "          21       0.91      0.81      0.86        90\n",
      "          22       0.99      0.86      0.92       120\n",
      "          23       0.95      0.99      0.97       150\n",
      "          24       0.99      0.98      0.98        90\n",
      "          25       0.94      0.99      0.97       480\n",
      "          26       0.90      0.83      0.86       180\n",
      "          27       0.89      0.55      0.68        60\n",
      "          28       0.95      0.98      0.97       150\n",
      "          29       0.98      1.00      0.99        90\n",
      "          30       0.92      0.88      0.90       150\n",
      "          31       0.99      0.98      0.99       270\n",
      "          32       0.88      1.00      0.94        60\n",
      "          33       0.95      1.00      0.97       210\n",
      "          34       0.92      0.99      0.96       120\n",
      "          35       0.99      0.98      0.99       390\n",
      "          36       0.99      0.99      0.99       120\n",
      "          37       0.97      1.00      0.98        60\n",
      "          38       0.98      0.99      0.99       690\n",
      "          39       0.91      0.98      0.94        90\n",
      "          40       0.88      0.90      0.89        90\n",
      "          41       0.98      0.80      0.88        60\n",
      "          42       1.00      0.99      0.99        90\n",
      "\n",
      "    accuracy                           0.97     12630\n",
      "   macro avg       0.96      0.95      0.95     12630\n",
      "weighted avg       0.97      0.97      0.97     12630\n",
      "\n"
     ]
    }
   ],
   "source": [
    "report = classification_report(y_test_classes, y_predicted_classes)\n",
    "print(report)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "20342ac5",
   "metadata": {},
   "source": [
    "Prikazacemo i vizuelno kako model radi."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "affd9ce9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x1080 with 12 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15,15))\n",
    "for i in range(0,12):\n",
    "    plt.subplot(4,3,i+1)\n",
    "    plt.imshow(X_test_gray_norm[i].squeeze(), cmap='gray', interpolation='none')\n",
    "    plt.title(\"Predicted {0}, True {1}\".format(y_predicted_classes[i], y_test_classes[i]))\n",
    "    plt.axis('off')\n",
    "plt.show() "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0650ae3d",
   "metadata": {},
   "source": [
    "Vidimo da su sve izdvojene slike dobro klasifikovane.\n",
    "\n",
    "Sada cemo pronaci one slike kod kojih se stvarna obelezja i predikcije razlikuju."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "5c58937c",
   "metadata": {},
   "outputs": [],
   "source": [
    "predicted_labels = np.argmax(y_predicted, axis=1)\n",
    "true_labels = np.argmax(y_test, axis=1)\n",
    "error_indexes = np.where(predicted_labels != true_labels)[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "bd9f85a5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Broj instanci koje je klasifikator pogrešno klasifikovao:  356\n"
     ]
    }
   ],
   "source": [
    "number_of_errors = len(error_indexes)\n",
    "print(\"Broj instanci koje je klasifikator pogrešno klasifikovao: \", number_of_errors)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "51fe7721",
   "metadata": {},
   "source": [
    "Prikazacemo neku slika sa pogresnom labelom."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "fa28ae06",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "error_index = np.random.choice(error_indexes)\n",
    "true_label = true_labels[error_index]\n",
    "predicted_label = predicted_labels[error_index]\n",
    "\n",
    "plt.title(\"Predicted {0}, True {1}\".format(true_label, predicted_label))\n",
    "plt.imshow(X_test_gray_norm[error_index].squeeze(), cmap='gray', interpolation='none')\n",
    "plt.axis('off')\n",
    "plt.show()"
   ]
  }
 ],
 "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": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}