cnn동물상분류.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "6fc5f4cc-244f-496a-8a06-9b436814b322",
   "metadata": {},
   "source": [
    "# 이미지 데이터 세트 구축"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "97191253-92ee-4903-b831-d5d723d09ec1",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "from sklearn.model_selection import train_test_split\n",
    "from PIL import Image\n",
    "import os, glob\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from tensorflow.keras.layers import Dense, Conv2D, MaxPooling2D, Flatten, Dropout\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.callbacks import ModelCheckpoint\n",
    "from keras.utils import np_utils\n",
    "from keras.models import load_model\n",
    "import matplotlib.pyplot as plt\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.applications import VGG16\n",
    "\n",
    "import warnings \n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "a01a74c8-3b0a-4081-a0e7-1d628970337c",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 데이터 경로 설정\n",
    "root_dir = \"data/\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "a485dd98-f2e9-466d-a646-1f15fd34b893",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 카테고리 지정 \n",
    "categories = [\"puppy\", \"cat\", \"dino\", \"rabbit\", \"fox\"]\n",
    "nb_classes = len(categories)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "0c162c8c-e776-4b09-99a9-eb6bf82969f5",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 이미지 크기 지정\n",
    "image_width = 64\n",
    "image_height = 64"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "686a4f49-3f87-45ba-80e2-a8821c0c6791",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "data/puppy/*.png\n",
      "data/cat/*.png\n",
      "data/dino/*.png\n",
      "data/rabbit/*.png\n",
      "data/fox/*.png\n"
     ]
    }
   ],
   "source": [
    "# 이미지 로딩\n",
    "X = [] # 이미지 데이터\n",
    "Y = [] # 레이블 데이터\n",
    "\n",
    "for idx, category in enumerate(categories):\n",
    "    image_dir = root_dir + category\n",
    "    files = glob.glob(image_dir + \"/\" + \"*.png\")\n",
    "    print(image_dir + \"/\" +\"*.png\")\n",
    "\n",
    "    for i, f in enumerate(files):\n",
    "        img = Image.open(f)\n",
    "        img = img.convert(\"RGB\")\n",
    "        img = img.resize((image_width, image_height))\n",
    "        data = np.asarray(img)\n",
    "        X.append(data)\n",
    "        Y.append(idx)\n",
    "\n",
    "X = np.array(X)\n",
    "Y = np.array(Y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "57e90e89-88ba-437d-8fd7-608164a30009",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train.shape : 36\n",
      "test.shape : 13\n"
     ]
    }
   ],
   "source": [
    "# 학습 데이터와 테스트 데이터 나누기\n",
    "x_train, x_test, y_train, y_test = train_test_split(X, Y, test_size=0.25, stratify=Y)\n",
    "xy = (x_train, x_test, y_train, y_test)\n",
    "print(\"train.shape :\", len(x_train))\n",
    "print(\"test.shape :\", len(x_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "6f38bad3-2b0e-4021-8315-0568e95fc7a4",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 데이터 파일 저장\n",
    "np.save(root_dir + \"animal.npy\", xy)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e29d3684-ff9c-4536-a4de-215cd9449deb",
   "metadata": {},
   "source": [
    "# 동물상 분류기 구축"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "7bb63c69-d4e1-488a-bd44-3653beb666f9",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 데이터 로드\n",
    "x_train, x_test, y_train, y_test = np.load(\"data/animal.npy\", allow_pickle=True)\n",
    "x_train = x_train.astype(\"float\") / 256\n",
    "x_test = x_test.astype(\"float\") / 256\n",
    "y_train = np_utils.to_categorical(y_train, nb_classes)\n",
    "y_test = np_utils.to_categorical(y_test, nb_classes)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "a3714485-fc4d-4cb2-bc5f-63860dbc16cd",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 모델 구축\n",
    "model = Sequential([\n",
    "                    Conv2D(32, (3,3), activation = 'relu', input_shape=x_train.shape[1:]),\n",
    "                    MaxPooling2D(2,2),\n",
    "                    Dropout(0.5),\n",
    "                    Conv2D(64,(3,3), activation = 'relu'),\n",
    "                    MaxPooling2D(2,2),\n",
    "                    Dropout(0.5),\n",
    "                    Flatten(),\n",
    "                    Dense(128, activation = 'relu'),\n",
    "                    Dropout(0.5),\n",
    "                    Dense(nb_classes, activation = 'softmax')\n",
    "                    ])"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7f3beb71-3be8-412b-a429-ad3e38b3487d",
   "metadata": {},
   "source": [
    "# VGG16 모델 구축\n",
    "transfer_model = VGG16(weights='imagenet', include_top=False, input_shape=(64, 64, 3))\n",
    "transfer_model.trainable=False\n",
    "model = Sequential([\n",
    "                    transfer_model,\n",
    "                    Flatten(),\n",
    "                    Dropout(0.5),\n",
    "                    Dense(512, activation='relu'),\n",
    "                    Dense(128, activation='relu'),\n",
    "                    Dense(5, activation='softmax'),\n",
    "])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "250411f6-8f90-4341-a1ef-e765c565439f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 모델 컴파일\n",
    "model.compile(loss='binary_crossentropy', \n",
    "              optimizer='rmsprop', \n",
    "              metrics=['acc'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "65130f70-03e8-4e5a-98cf-6f1702f3197b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 체크포인트 설정\n",
    "checkpoint_path = \"my_checkpoint.ckpt\"\n",
    "checkpoint = ModelCheckpoint(filepath=checkpoint_path, \n",
    "                             save_weights_only=True, \n",
    "                             save_best_only=True, \n",
    "                             monitor='val_loss', \n",
    "                             verbose=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "ee5d0fdf-182a-44f1-ab1d-810f714bc9b3",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 모델 학습\n",
    "history = model.fit(x_train, y_train,\n",
    "                    validation_data=(x_test, y_test),\n",
    "                    batch_size=32,\n",
    "                    epochs=50,\n",
    "                    verbose=0,\n",
    "                    callbacks=[checkpoint])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "079d8a64-7b3a-4632-ae32-65d53537f58f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 16ms/step - loss: 0.4483 - acc: 0.4615\n"
     ]
    },
    {
     "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": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x29f8e7a6f10>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "# 모델 평가\n",
    "model.load_weights(checkpoint_path)\n",
    "model.evaluate(x_test, y_test)\n",
    "\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "acc = history.history['acc']\n",
    "val_acc = history.history['val_acc']\n",
    "\n",
    "plt.plot(loss)\n",
    "plt.plot(val_loss, 'r')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.title('Loss')\n",
    "plt.legend([\"Training\", \"Validation\"])\n",
    "\n",
    "plt.show()\n",
    "\n",
    "plt.plot(acc)\n",
    "plt.plot(val_acc, 'r')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.title('Accuracy')\n",
    "plt.legend(['Training', 'Validation'], loc=4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "18244c98-82ab-4873-b24d-2d97f2167f38",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 모델 저장\n",
    "model.save(\"model/animal_model.h5\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6cd8f27a-d747-41e2-b211-c16a41a9cafe",
   "metadata": {},
   "source": [
    "# 모델 테스트"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2fc52dcf-e086-489c-bb9b-ec26c7fa37fb",
   "metadata": {},
   "source": [
    "- 남성 유저들의 이미지를 활용하여 동물상을 분류"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "ef5c2e1f-2df2-442f-8132-169d598df801",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 테스트 데이터 전처리\n",
    "df = pd.read_csv('data/member_mate_sample.csv')\n",
    "df1 = df.copy()\n",
    "\n",
    "df = df[['mem_no','rprsn_photo','mem_sex']]\n",
    "df.dropna(inplace=True)\n",
    "df['rprsn_photo'] = df['rprsn_photo'].str[:-9].str.replace('|','')\n",
    "\n",
    "path = 'data/image/'\n",
    "files = os.listdir(path)\n",
    "\n",
    "data2 = pd.DataFrame()\n",
    "data2['a'] = files\n",
    "\n",
    "data2['mem_no'] = data2['a'].str.split('_').str[0]\n",
    "data2['mem_no'] = data2['mem_no'].astype('int64')\n",
    "\n",
    "data3 = pd.merge(data2, df1[['mem_no','mem_sex']])\n",
    "\n",
    "m_image = data3[data3['mem_sex'] == 'm']\n",
    "m_image.reset_index(inplace=True, drop=True)\n",
    "\n",
    "root_dir = \"data/image/\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "421aaa96-9d54-443d-a7eb-50a7190a87af",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 테스트 데이터 구축\n",
    "image_files = []\n",
    "\n",
    "for i in range(len(m_image)):\n",
    "    image_files.append(root_dir + m_image['a'][i])\n",
    "    \n",
    "image_size = 64\n",
    "categories = [\"puppy\", \"cat\", \"dino\", \"rabbit\", \"fox\"]\n",
    "nb_classes = len(categories)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "c5c7ba6b-6b41-4dd0-ab7f-1d0919fdf02e",
   "metadata": {},
   "outputs": [],
   "source": [
    "X = []\n",
    "files = []\n",
    "for fname in image_files:\n",
    "    img = Image.open(fname)\n",
    "    img = img.convert(\"RGB\")\n",
    "    img = img.resize((image_size, image_size))\n",
    "    in_data = np.asarray(img)\n",
    "    in_data = in_data.astype(\"float\") / 256\n",
    "    X.append(in_data)\n",
    "    files.append(fname)\n",
    "\n",
    "X = np.array(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "2d17337d-8592-4ad5-a0d6-c3b2c4ab3b5e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 모댈 불러오기\n",
    "model = load_model('model/animal_model.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "ce4f20e2-bc94-45d8-9e4b-9ccaa8170a4d",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 예측\n",
    "pred = model.predict(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "7c20c1f0-38c1-47f9-ac99-689df70a3d8e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# 예측 데이터셋 생성\n",
    "animal_list = []\n",
    "\n",
    "for i in range(len(pred)):\n",
    "    animal_list.append(categories[pred[i].argmax()])\n",
    "    \n",
    "m_image['animal'] = animal_list\n",
    "m_image = m_image[['mem_no','animal']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "85f9730d-d71e-43dd-951a-06652a38fb1a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "cat       6074\n",
       "dino      1476\n",
       "fox       1086\n",
       "rabbit     457\n",
       "puppy      215\n",
       "Name: animal, dtype: int64"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "m_image['animal'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "e4730a4b-cbaf-403f-b4c1-a416f921310f",
   "metadata": {},
   "outputs": [],
   "source": [
    "# to_csv\n",
    "m_image.to_csv('data/animal.csv', index=False)"
   ]
  }
 ],
 "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
}