02_evaluate.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 🚩 Step:2-1\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "import tensorflow as tf\n",
    "import tensorflow.keras as keras\n",
    "import extra_keras_datasets.emnist as emnist\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 🚩 Step:2-2\n",
    "# EMNIST-LettersのデータをNumPy形式で取得（評価用のデータだけを取得）\n",
    "_, (v_images, v_labels) = emnist.load_data(type=\"letters\")\n",
    "\n",
    "# データ形式を4次元データ (batch, height, width, channel)に変換\n",
    "# 各ピクセルの情報も0..255から0..1の範囲にノーマライズ\n",
    "def adjust_image(img):\n",
    "    img = img.astype(\"float32\")\n",
    "    img = img / 255.0\n",
    "    img = img.reshape(-1, 28, 28, 1)  # 3D→4D\n",
    "    return img\n",
    "\n",
    "\n",
    "v_images = adjust_image(v_images)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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)        320       \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 28, 28, 32)        9248      \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 28, 28, 32)        25632     \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 14, 14, 32)        0         \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 14, 14, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 14, 14, 64)        18496     \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 14, 14, 64)        36928     \n",
      "_________________________________________________________________\n",
      "conv2d_5 (Conv2D)            (None, 14, 14, 64)        102464    \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 7, 7, 64)          0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 7, 7, 64)          0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 3136)              0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 128)               401536    \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 27)                3483      \n",
      "=================================================================\n",
      "Total params: 598,107\n",
      "Trainable params: 598,107\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# 🚩 Step:2-3\n",
    "# 学習済みモデルの読み込み\n",
    "model = keras.models.load_model(\"data/emnist-model.h5\")\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "650/650 [==============================] - 1s 1ms/step - loss: 0.1547 - accuracy: 0.9516\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.15471120178699493, 0.9516345858573914]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 🚩 Step:2-4\n",
    "# 評価用データを使ってモデルの評価\n",
    "model.evaluate(v_images, v_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[2.5993692e-30 1.0000000e+00 1.4818017e-12 9.3399866e-18 1.3195496e-11\n",
      " 1.9532371e-13 4.6664905e-11 1.4196876e-08 1.6139170e-08 3.0195570e-15\n",
      " 6.1250185e-15 2.9840820e-15 6.5325557e-12 2.8742930e-12 1.5786698e-12\n",
      " 1.4525284e-16 6.1413735e-11 1.4041309e-11 1.1188976e-09 1.6182151e-14\n",
      " 6.7944023e-16 2.8441763e-21 3.0535916e-21 9.3609433e-14 6.3657008e-17\n",
      " 1.4761984e-13 1.6906218e-16]\n"
     ]
    }
   ],
   "source": [
    "# 🚩 Step:2-5\n",
    "# 予測\n",
    "predictions = model.predict(v_images)\n",
    "\n",
    "# 予測結果の一部を確認\n",
    "print(predictions[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x216 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x216 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x216 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "class_names = list(\" ABCDEFGHIJKLMNOPQRSTUVWXYZ\")\n",
    "assert len(class_names) == 27  # EMNIST-Letters is 27-class\n",
    "\n",
    "\n",
    "# Matplotlib 3.3.0より前のimshow互換用 (h,w,1)→(h,w)\n",
    "def compatible_image(img):\n",
    "    if img.ndim == 3 and img.shape[2] == 1:\n",
    "        img = img.reshape(img.shape[0], img.shape[1])\n",
    "    return img\n",
    "\n",
    "\n",
    "# 予測結果を分かりやすく表示\n",
    "def show_prediction(i, predictions, images, true_labels):\n",
    "    prediction, image, true_label = predictions[i], images[i], true_labels[i]\n",
    "    predicted_label = np.argmax(prediction)\n",
    "\n",
    "    plt.rcParams[\"font.size\"] = 10\n",
    "    fig, (ax1, ax2) = plt.subplots(\n",
    "        1, 2, figsize=(10, 3), gridspec_kw={\"width_ratios\": [1, 5]}\n",
    "    )\n",
    "\n",
    "    label = \"True: {}\\nPredicted: {}\\nCorrect: {}\".format(\n",
    "        class_names[true_label],\n",
    "        class_names[predicted_label],\n",
    "        true_label == predicted_label,\n",
    "    )\n",
    "    ax1.set_title(label, loc=\"left\", fontsize=12)\n",
    "    ax1.set_anchor(\"S\")  # S=Bottom\n",
    "    ax1.set_xticks([])\n",
    "    ax1.set_yticks([])\n",
    "    ax1.imshow(compatible_image(image), cmap=plt.cm.binary)\n",
    "\n",
    "    x_pos = np.arange(len(class_names))\n",
    "    container = ax2.bar(x_pos, prediction, color=\"#999\")\n",
    "    container[predicted_label].set_color(\"#C00\")  # 予測\n",
    "    container[true_label].set_color(\"#0A0\")  # 正解\n",
    "\n",
    "    ax2.set_yticks(np.arange(0, 1.1, 0.1))\n",
    "    ax2.set_xticks(x_pos)\n",
    "    ax2.set_xticklabels(class_names)\n",
    "    ax2.grid(True, axis=\"y\")\n",
    "    ax2.set_axisbelow(True)\n",
    "\n",
    "    plt.tight_layout()\n",
    "    plt.show()\n",
    "\n",
    "\n",
    "# 🚩 Step:2-6\n",
    "show_prediction(1, predictions, v_images, v_labels)\n",
    "show_prediction(1000, predictions, v_images, v_labels)  # 複数候補が出る\n",
    "show_prediction(5480, predictions, v_images, v_labels)  # 正解ラベルが微妙"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def show_confusion_matrix(predictions, true_labels, class_names):\n",
    "    pred_labels = np.argmax(predictions, axis=1)\n",
    "    cm = tf.math.confusion_matrix(true_labels, pred_labels)\n",
    "    cm = cm / cm.numpy().sum(axis=1)[:, tf.newaxis]  # Normalize\n",
    "    cm = np.nan_to_num(cm)\n",
    "\n",
    "    plt.rcParams[\"font.size\"] = 10\n",
    "    plt.figure(figsize=(12, 10))\n",
    "    sns.heatmap(\n",
    "        cm,\n",
    "        vmin=0.0,\n",
    "        vmax=1.0,\n",
    "        cmap=\"Blues\",\n",
    "        cbar_kws={\"aspect\": 50},\n",
    "        square=True,\n",
    "        linewidths=1,\n",
    "        xticklabels=class_names,\n",
    "        yticklabels=class_names,\n",
    "    )\n",
    "    plt.show()\n",
    "\n",
    "\n",
    "# 🚩 Step:2-7\n",
    "# データ全体での結果を表示（混同行列：縦軸=正解ラベル、横軸=予測ラベル）\n",
    "show_confusion_matrix(predictions, v_labels, class_names)"
   ]
  }
 ],
 "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": "3.8.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}