{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "d5jFanC8NPTN"
   },
   "outputs": [],
   "source": [
    "import os\n",
    "os.chdir('/content/drive/MyDrive/Colab Notebooks/is_dnn/labworks/LW1')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "H39A4nsqNuxn"
   },
   "outputs": [],
   "source": [
    "# импорт модулей\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from keras.datasets import mnist\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.utils import to_categorical\n",
    "from sklearn.model_selection import train_test_split\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "from PIL import Image\n",
    "import os\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 462,
     "status": "ok",
     "timestamp": 1759127274975,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "iNenKkcoRXrs",
    "outputId": "041dc403-e177-4f0d-edbb-40902c8954fd"
   },
   "outputs": [],
   "source": [
    "# Загрузка датасета\n",
    "(X_train_orig, y_train_orig), (X_test_orig, y_test_orig) = mnist.load_data()\n",
    "\n",
    "# разбиваем выборку на обучающую и тестовую выборку\n",
    "X = np.concatenate((X_train_orig, X_test_orig))\n",
    "y = np.concatenate((y_train_orig, y_test_orig))\n",
    "\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    X, y,\n",
    "    test_size=10000,\n",
    "    train_size=60000,\n",
    "    random_state=3,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 224
    },
    "executionInfo": {
     "elapsed": 243,
     "status": "ok",
     "timestamp": 1759127328545,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "bt-EmlYARsCL",
    "outputId": "611d9110-39a2-46dd-94ce-b0fea7005b87"
   },
   "outputs": [],
   "source": [
    "# Вывод первых 4 изображений\n",
    "fig, axes = plt.subplots(1, 4, figsize=(12, 3))\n",
    "for i in range(4):\n",
    "    axes[i].imshow(X_train[i], cmap='gray')\n",
    "    axes[i].set_title(f'Метка: {y_train[i]}')\n",
    "    axes[i].axis('off')\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 134,
     "status": "ok",
     "timestamp": 1759127329383,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "5WUu3_97TjSa",
    "outputId": "7a75ca25-58fe-447d-8bef-547159e6479d"
   },
   "outputs": [],
   "source": [
    "# развернем каждое изображение 28*28 в вектор 784\n",
    "num_pixels = X_train.shape[1] * X_train.shape[2]\n",
    "X_train = X_train.reshape(X_train.shape[0], num_pixels) / 255\n",
    "X_test = X_test.reshape(X_test.shape[0], num_pixels) / 255\n",
    "print('Shape of transformed X train:', X_train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 7,
     "status": "ok",
     "timestamp": 1759127330016,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "sUJfDepgUauM",
    "outputId": "2a9bc3d0-8bdb-4971-f3d4-ca21cb14fca6"
   },
   "outputs": [],
   "source": [
    "# переведем метки в one-hot\n",
    "y_train = to_categorical(y_train)\n",
    "y_test = to_categorical(y_test)\n",
    "print('Shape of transformed y train:', y_train.shape)\n",
    "num_classes = y_train.shape[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "executionInfo": {
     "elapsed": 207500,
     "status": "ok",
     "timestamp": 1759127540337,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "f3UzOyf_V2HQ",
    "outputId": "9ffc63df-23bf-4947-f3c0-2a3e507034f1"
   },
   "outputs": [],
   "source": [
    "model_0 = Sequential()\n",
    "model_0.add(Dense(units=num_classes, input_dim=num_pixels, activation='softmax'))\n",
    "\n",
    "# Компиляция модели\n",
    "model_0.compile(loss='categorical_crossentropy',\n",
    "                optimizer='sgd',\n",
    "                metrics=['accuracy'])\n",
    "\n",
    "# Вывод информации об архитектуре\n",
    "print(\"Архитектура однослойной сети:\")\n",
    "model_0.summary()\n",
    "\n",
    "# Обучение модели\n",
    "history_0 = model_0.fit(X_train, y_train,\n",
    "                        validation_split=0.1,\n",
    "                        epochs=50)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 533
    },
    "executionInfo": {
     "elapsed": 260,
     "status": "ok",
     "timestamp": 1759127542723,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "5yjxVnmFmpbV",
    "outputId": "b182a139-da24-491e-9ebd-7ba1e39eec84"
   },
   "outputs": [],
   "source": [
    "# График функции ошибки по эпохам\n",
    "plt.figure(figsize=(10, 6))\n",
    "plt.plot(history_0.history['loss'], label='Обучающая выборка')\n",
    "plt.plot(history_0.history['val_loss'], label='Валидационная выборка')\n",
    "plt.title('Функция ошибки по эпохам (Однослойная сеть)')\n",
    "plt.xlabel('Эпохи')\n",
    "plt.ylabel('Ошибка')\n",
    "plt.legend()\n",
    "plt.grid(True)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 1058,
     "status": "ok",
     "timestamp": 1759127544719,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "NF_SsO8wiEUT",
    "outputId": "ffe554c2-1c94-42b7-cc63-8f407418c4ce"
   },
   "outputs": [],
   "source": [
    "# Оценка на тестовых данных\n",
    "scores_0 = model_0.evaluate(X_test, y_test, verbose=0)\n",
    "print(\"Результаты однослойной сети:\")\n",
    "print(f\"Ошибка на тестовых данных: {scores_0[0]}\")\n",
    "print(f\"Точность на тестовых данных: {scores_0[1]}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "fFcoQGdPnDFx"
   },
   "outputs": [],
   "source": [
    "# Функция для создания и обучения модели\n",
    "def create_and_train_model(hidden_units, model_name):\n",
    "    model = Sequential()\n",
    "    model.add(Dense(units=hidden_units, input_dim=num_pixels, activation='sigmoid'))\n",
    "    model.add(Dense(units=num_classes, activation='softmax'))\n",
    "\n",
    "    model.compile(loss='categorical_crossentropy',\n",
    "                  optimizer='sgd',\n",
    "                  metrics=['accuracy'])\n",
    "\n",
    "    history = model.fit(X_train, y_train,\n",
    "                        validation_split=0.1,\n",
    "                        epochs=50)\n",
    "\n",
    "    scores = model.evaluate(X_test, y_test, verbose=0)\n",
    "\n",
    "    return model, history, scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "XPlFkV40joAT"
   },
   "outputs": [],
   "source": [
    "# Эксперименты с разным количеством нейронов\n",
    "hidden_units_list = [100, 300, 500]\n",
    "models_1 = {}\n",
    "histories_1 = {}\n",
    "scores_1 = {}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 638747,
     "status": "ok",
     "timestamp": 1759128186868,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "je0i_8HxjvpB",
    "outputId": "5cb19edc-9162-4c23-92d8-1671e62b2bb3"
   },
   "outputs": [],
   "source": [
    "# Обучение сетей с одним скрытым слоем\n",
    "for units in hidden_units_list:\n",
    "    print(f\"\\nОбучение модели с {units} нейронами...\")\n",
    "    model, history, scores = create_and_train_model(units, f\"model_{units}\")\n",
    "\n",
    "    models_1[units] = model\n",
    "    histories_1[units] = history\n",
    "    scores_1[units] = scores\n",
    "\n",
    "    print(f\"Точность: {scores[1]}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 20,
     "status": "ok",
     "timestamp": 1759128269112,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "TbbBBxeMmy6c",
    "outputId": "08b5a164-1e62-456c-80b9-bc7247be3fd3"
   },
   "outputs": [],
   "source": [
    "# Выбор наилучшей модели\n",
    "best_units_1 = max(scores_1.items(), key=lambda x: x[1][1])[0]\n",
    "print(f\"\\nНаилучшее количество нейронов: {best_units_1}\")\n",
    "print(f\"Точность: {scores_1[best_units_1][1]}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 279
    },
    "executionInfo": {
     "elapsed": 620,
     "status": "ok",
     "timestamp": 1759128272502,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "EFEBxB2qm1fF",
    "outputId": "d436ac7a-e33b-4cc2-e324-5decfc29de94"
   },
   "outputs": [],
   "source": [
    "# Графики ошибок для всех моделей\n",
    "plt.figure(figsize=(15, 5))\n",
    "for i, units in enumerate(hidden_units_list, 1):\n",
    "    plt.subplot(1, 3, i)\n",
    "    plt.plot(histories_1[units].history['loss'], label='Обучающая')\n",
    "    plt.plot(histories_1[units].history['val_loss'], label='Валидационная')\n",
    "    plt.title(f'{units} нейронов')\n",
    "    plt.xlabel('Эпохи')\n",
    "    plt.ylabel('Ошибка')\n",
    "    plt.legend()\n",
    "    plt.grid(True)\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "EJBwT5vhnnSG"
   },
   "outputs": [],
   "source": [
    "# Добавление второго скрытого слоя\n",
    "second_layer_units = [50, 100]\n",
    "models_2 = {}\n",
    "histories_2 = {}\n",
    "scores_2 = {}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 420194,
     "status": "ok",
     "timestamp": 1759128698216,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "K5vweMySno3t",
    "outputId": "b67b155c-89ea-46e0-a590-a588433e0a38"
   },
   "outputs": [],
   "source": [
    "for units_2 in second_layer_units:\n",
    "    print(f\"\\nОбучение модели со вторым слоем {units_2} нейронов\")\n",
    "\n",
    "    model = Sequential()\n",
    "    model.add(Dense(units=best_units_1, input_dim=num_pixels, activation='sigmoid'))\n",
    "    model.add(Dense(units=units_2, activation='sigmoid'))\n",
    "    model.add(Dense(units=num_classes, activation='softmax'))\n",
    "\n",
    "    model.compile(loss='categorical_crossentropy',\n",
    "                  optimizer='sgd',\n",
    "                  metrics=['accuracy'])\n",
    "\n",
    "    history = model.fit(X_train, y_train,\n",
    "                        validation_split=0.1,\n",
    "                        epochs=50)\n",
    "\n",
    "    scores = model.evaluate(X_test, y_test)\n",
    "\n",
    "    models_2[units_2] = model\n",
    "    histories_2[units_2] = history\n",
    "    scores_2[units_2] = scores\n",
    "\n",
    "    print(f\"Точность: {scores[1]}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 14,
     "status": "ok",
     "timestamp": 1759129484222,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "9lJtmn_oSVkB",
    "outputId": "b49d6a95-574a-4ce5-e23f-eea49ba83603"
   },
   "outputs": [],
   "source": [
    "# Выбор наилучшей двухслойной модели\n",
    "best_units_2 = max(scores_2.items(), key=lambda x: x[1][1])[0]\n",
    "print(f\"\\nНаилучшее количество нейронов во втором слое: {best_units_2}\")\n",
    "print(f\"Точность: {scores_2[best_units_2][1]:.4f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "p0aeriYzShJk"
   },
   "outputs": [],
   "source": [
    "# Сбор результатов\n",
    "results = {\n",
    "    '0 слоев': {'нейроны_1': '-', 'нейроны_2': '-', 'точность': scores_0[1]},\n",
    "    '1 слой_100': {'нейроны_1': 100, 'нейроны_2': '-', 'точность': scores_1[100][1]},\n",
    "    '1 слой_300': {'нейроны_1': 300, 'нейроны_2': '-', 'точность': scores_1[300][1]},\n",
    "    '1 слой_500': {'нейроны_1': 500, 'нейроны_2': '-', 'точность': scores_1[500][1]},\n",
    "    '2 слоя_50': {'нейроны_1': best_units_1, 'нейроны_2': 50, 'точность': scores_2[50][1]},\n",
    "    '2 слоя_100': {'нейроны_1': best_units_1, 'нейроны_2': 100, 'точность': scores_2[100][1]}\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 33,
     "status": "ok",
     "timestamp": 1759130442386,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "SHr6z7jbSmOG",
    "outputId": "2d40f526-7756-4554-f110-2242e34e58a0"
   },
   "outputs": [],
   "source": [
    "# Создаем DataFrame из результатов\n",
    "df_results = pd.DataFrame([\n",
    "    {'Кол-во скрытых слоев': 0, 'Нейроны_1_слоя': '-', 'Нейроны_2_слоя': '-', 'Точность': results['0 слоев']['точность']},\n",
    "    {'Кол-во скрытых слоев': 1, 'Нейроны_1_слоя': 100, 'Нейроны_2_слоя': '-', 'Точность': results['1 слой_100']['точность']},\n",
    "    {'Кол-во скрытых слоев': 1, 'Нейроны_1_слоя': 300, 'Нейроны_2_слоя': '-', 'Точность': results['1 слой_300']['точность']},\n",
    "    {'Кол-во скрытых слоев': 1, 'Нейроны_1_слоя': 500, 'Нейроны_2_слоя': '-', 'Точность': results['1 слой_500']['точность']},\n",
    "    {'Кол-во скрытых слоев': 2, 'Нейроны_1_слоя': best_units_1, 'Нейроны_2_слоя': 50, 'Точность': results['2 слоя_50']['точность']},\n",
    "    {'Кол-во скрытых слоев': 2, 'Нейроны_1_слоя': best_units_1, 'Нейроны_2_слоя': 100, 'Точность': results['2 слоя_100']['точность']}\n",
    "])\n",
    "\n",
    "print(\" \" * 20 + \"ТАБЛИЦА РЕЗУЛЬТАТОВ\")\n",
    "print(\"=\" * 70)\n",
    "# print(df_results.to_string(index=False, formatters={\n",
    "#     'Точность': '{:.4f}'.format\n",
    "# }))\n",
    "print(df_results.reset_index(drop=True))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "executionInfo": {
     "elapsed": 41,
     "status": "ok",
     "timestamp": 1759130490602,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "PTC5CUJeWQ_V",
    "outputId": "e8009546-5876-4427-9815-9aac53db8593"
   },
   "outputs": [],
   "source": [
    "# Выбор наилучшей модели\n",
    "best_model_type = max(results.items(), key=lambda x: x[1]['точность'])[0]\n",
    "best_accuracy = results[best_model_type]['точность']\n",
    "print(f\"\\nНаилучшая архитектура: {best_model_type}\")\n",
    "print(f\"Точность: {best_accuracy:.4f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "JRPRpppHWV8-"
   },
   "outputs": [],
   "source": [
    "# Определение наилучшей модели\n",
    "if '0' in best_model_type:\n",
    "    best_model = model_0\n",
    "elif '1' in best_model_type:\n",
    "    best_neurons = int(best_model_type.split('_')[1])\n",
    "    best_model = models_1[best_neurons]\n",
    "else:\n",
    "    best_neurons_2 = int(best_model_type.split('_')[1])\n",
    "    best_model = models_2[best_neurons_2]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "wIlYoP_HSFph"
   },
   "outputs": [],
   "source": [
    "# Сохранение модели\n",
    "best_model.save('best_mnist_model.keras')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 517
    },
    "executionInfo": {
     "elapsed": 178,
     "status": "ok",
     "timestamp": 1759132236751,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "Kh6-u-8OoHny",
    "outputId": "7c89ce31-3967-41bb-ff79-6e77e7f5d19b"
   },
   "outputs": [],
   "source": [
    "# вывод тестового изображения и результата распознавания (1)\n",
    "n = 123\n",
    "result = best_model.predict(X_test[n:n+1])\n",
    "print('NN output:', result)\n",
    "plt.imshow(X_test[n].reshape(28,28), cmap=plt.get_cmap('gray'))\n",
    "plt.show()\n",
    "print('Real mark: ', str(np.argmax(y_test[n])))\n",
    "print('NN answer: ', str(np.argmax(result)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 517
    },
    "executionInfo": {
     "elapsed": 284,
     "status": "ok",
     "timestamp": 1759132262259,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "WVGJRtVZc8V-",
    "outputId": "5d4ed790-d7ce-4569-d667-49733f75e3cb"
   },
   "outputs": [],
   "source": [
    "# вывод тестового изображения и результата распознавания (3)\n",
    "n = 353\n",
    "result = best_model.predict(X_test[n:n+1])\n",
    "print('NN output:', result)\n",
    "plt.imshow(X_test[n].reshape(28,28), cmap=plt.get_cmap('gray'))\n",
    "plt.show()\n",
    "print('Real mark: ', str(np.argmax(y_test[n])))\n",
    "print('NN answer: ', str(np.argmax(result)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "z5-YYw4uosUB"
   },
   "outputs": [],
   "source": [
    "# загрузка собственного изображения (Цифры 2 и 7)\n",
    "from PIL import Image\n",
    "file_data_2 = Image.open('2.png')\n",
    "file_data_7 = Image.open('7.png')\n",
    "file_data_2 = file_data_2.convert('L') # перевод в градации серого\n",
    "file_data_7 = file_data_7.convert('L') # перевод в градации серого\n",
    "test_img_2 = np.array(file_data_2)\n",
    "test_img_7 = np.array(file_data_7)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 465
    },
    "executionInfo": {
     "elapsed": 156,
     "status": "ok",
     "timestamp": 1759130765327,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "dv17bJVVuslg",
    "outputId": "d9b5b55c-75d9-4180-f93c-22befad0633c"
   },
   "outputs": [],
   "source": [
    "# вывод собственного изображения (цифра 2)\n",
    "plt.imshow(test_img_2, cmap=plt.get_cmap('gray'))\n",
    "plt.show()\n",
    "# предобработка\n",
    "test_img_2 = test_img_2 / 255\n",
    "test_img_2 = test_img_2.reshape(1, num_pixels)\n",
    "# распознавание\n",
    "result = best_model.predict(test_img_2)\n",
    "print('I think it\\'s ', np.argmax(result))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 465
    },
    "executionInfo": {
     "elapsed": 651,
     "status": "ok",
     "timestamp": 1759130799101,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "rhNzATtGuxbD",
    "outputId": "11cd1569-d370-4070-c8de-02035699d8bb"
   },
   "outputs": [],
   "source": [
    "# вывод собственного изображения (цифра 7)\n",
    "plt.imshow(test_img_7, cmap=plt.get_cmap('gray'))\n",
    "plt.show()\n",
    "# предобработка\n",
    "test_img_7 = test_img_7 / 255\n",
    "test_img_7 = test_img_7.reshape(1, num_pixels)\n",
    "# распознавание\n",
    "result = best_model.predict(test_img_7)\n",
    "print('I think it\\'s ', np.argmax(result))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "4rIUmwfYXcgh"
   },
   "outputs": [],
   "source": [
    "# Тестирование на собственных повернутых изображениях\n",
    "file_data_2_90 = Image.open('2_90.png')\n",
    "file_data_7_90 = Image.open('7_90.png')\n",
    "file_data_2_90 = file_data_2_90.convert('L') # перевод в градации серого\n",
    "file_data_7_90 = file_data_7_90.convert('L') # перевод в градации серого\n",
    "test_img_2_90 = np.array(file_data_2_90)\n",
    "test_img_7_90= np.array(file_data_7_90)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 465
    },
    "executionInfo": {
     "elapsed": 445,
     "status": "ok",
     "timestamp": 1759131775554,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "xeYNkU0OZqg2",
    "outputId": "b17a7b99-ce57-45fa-c069-74ea8374c3d3"
   },
   "outputs": [],
   "source": [
    "# вывод собственного изображения (цифра 2)\n",
    "plt.imshow(test_img_2_90, cmap=plt.get_cmap('gray'))\n",
    "plt.show()\n",
    "# предобработка\n",
    "test_img_2_90 = test_img_2_90 / 255\n",
    "test_img_2_90 = test_img_2_90.reshape(1, num_pixels)\n",
    "# распознавание\n",
    "result = best_model.predict(test_img_2_90)\n",
    "print('I think it\\'s ', np.argmax(result))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 465
    },
    "executionInfo": {
     "elapsed": 238,
     "status": "ok",
     "timestamp": 1759131800104,
     "user": {
      "displayName": "Legal People",
      "userId": "00818738730090246603"
     },
     "user_tz": -180
    },
    "id": "8JZajicXbNSA",
    "outputId": "016e8c12-472d-4a15-c420-cd955edef901"
   },
   "outputs": [],
   "source": [
    "# вывод собственного изображения (цифра 7)\n",
    "plt.imshow(test_img_7_90, cmap=plt.get_cmap('gray'))\n",
    "plt.show()\n",
    "# предобработка\n",
    "test_img_7_90 = test_img_7_90 / 255\n",
    "test_img_7_90 = test_img_7_90.reshape(1, num_pixels)\n",
    "# распознавание\n",
    "result = best_model.predict(test_img_7_90)\n",
    "print('I think it\\'s ', np.argmax(result))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "DQJNpFTOZ7Z6"
   },
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "gpuType": "T4",
   "provenance": [
    {
     "file_id": "1HorM0jtoJfNfcuh_uXJu8ODaEJ6xOUu-",
     "timestamp": 1759209370437
    }
   ]
  },
  "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": "3.11.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}