5.2.2 卷积神经网络-猫狗分类-基准模型.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "先训练一个简单的卷积神经网络，不做任何正则化，后面将使用数据增强、特征提取、微调等处理来优化模型。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 2000 images belonging to 2 classes.\n",
      "Found 1000 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "# 数据预处理\n",
    "# 快速创建python生成器\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "\n",
    "train_dir = './data//cats_and_dogs_small/train'\n",
    "validation_dir = './data//cats_and_dogs_small/validation'\n",
    "\n",
    "train_datagen = ImageDataGenerator(rescale=1./255)   \n",
    "test_datagen = ImageDataGenerator(rescale=1./255) \n",
    "train_generator = train_datagen.flow_from_directory( \n",
    "        train_dir,   \n",
    "        target_size=(150, 150),   \n",
    "        batch_size=20, \n",
    "        class_mode='binary')   \n",
    "validation_generator = test_datagen.flow_from_directory(\n",
    "        validation_dir, \n",
    "        target_size=(150, 150), \n",
    "        batch_size=20, \n",
    "        class_mode='binary')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/fc/anaconda3/lib/python3.6/site-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n"
     ]
    }
   ],
   "source": [
    "# 构建网络\n",
    "from keras import layers \n",
    "from keras import models \n",
    "model = models.Sequential() \n",
    "model.add(layers.Conv2D(32, (3, 3), activation='relu',\n",
    "                        input_shape=(150, 150, 3))) \n",
    "model.add(layers.MaxPooling2D((2, 2))) \n",
    "model.add(layers.Conv2D(64, (3, 3), activation='relu'))\n",
    "model.add(layers.MaxPooling2D((2, 2))) \n",
    "model.add(layers.Conv2D(128, (3, 3), activation='relu'))\n",
    "model.add(layers.MaxPooling2D((2, 2))) \n",
    "model.add(layers.Conv2D(128, (3, 3), activation='relu'))\n",
    "model.add(layers.MaxPooling2D((2, 2))) \n",
    "\n",
    "model.add(layers.Flatten()) \n",
    "model.add(layers.Dense(512, activation='relu')) \n",
    "model.add(layers.Dense(1, activation='sigmoid'))\n",
    "\n",
    "from keras import optimizers \n",
    "model.compile(loss='binary_crossentropy', \n",
    "              optimizer=optimizers.RMSprop(lr=1e-4), \n",
    "              metrics=['acc'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/fc/anaconda3/lib/python3.6/site-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.cast instead.\n",
      "Epoch 1/30\n",
      "100/100 [==============================] - 132s 1s/step - loss: 0.6851 - acc: 0.5455 - val_loss: 0.6722 - val_acc: 0.5750\n",
      "Epoch 2/30\n",
      "100/100 [==============================] - 115s 1s/step - loss: 0.6438 - acc: 0.6260 - val_loss: 0.6216 - val_acc: 0.6600\n",
      "Epoch 3/30\n",
      "100/100 [==============================] - 121s 1s/step - loss: 0.5925 - acc: 0.6860 - val_loss: 0.6029 - val_acc: 0.6620\n",
      "Epoch 4/30\n",
      "100/100 [==============================] - 116s 1s/step - loss: 0.5528 - acc: 0.7155 - val_loss: 0.5983 - val_acc: 0.6710\n",
      "Epoch 5/30\n",
      "100/100 [==============================] - 119s 1s/step - loss: 0.5264 - acc: 0.7275 - val_loss: 0.5765 - val_acc: 0.6950\n",
      "Epoch 6/30\n",
      "100/100 [==============================] - 117s 1s/step - loss: 0.5016 - acc: 0.7505 - val_loss: 0.5717 - val_acc: 0.6960\n",
      "Epoch 7/30\n",
      "100/100 [==============================] - 115s 1s/step - loss: 0.4748 - acc: 0.7675 - val_loss: 0.5674 - val_acc: 0.7050\n",
      "Epoch 8/30\n",
      "100/100 [==============================] - 115s 1s/step - loss: 0.4551 - acc: 0.7835 - val_loss: 0.5808 - val_acc: 0.7110\n",
      "Epoch 9/30\n",
      "100/100 [==============================] - 116s 1s/step - loss: 0.4362 - acc: 0.7990 - val_loss: 0.5615 - val_acc: 0.7140\n",
      "Epoch 10/30\n",
      "100/100 [==============================] - 112s 1s/step - loss: 0.4032 - acc: 0.8110 - val_loss: 0.5546 - val_acc: 0.7340\n",
      "Epoch 11/30\n",
      "100/100 [==============================] - 115s 1s/step - loss: 0.3769 - acc: 0.8365 - val_loss: 0.5451 - val_acc: 0.7340\n",
      "Epoch 12/30\n",
      "100/100 [==============================] - 113s 1s/step - loss: 0.3543 - acc: 0.8470 - val_loss: 0.5511 - val_acc: 0.7310\n",
      "Epoch 13/30\n",
      "100/100 [==============================] - 115s 1s/step - loss: 0.3241 - acc: 0.8610 - val_loss: 0.5670 - val_acc: 0.7330\n",
      "Epoch 14/30\n",
      "100/100 [==============================] - 117s 1s/step - loss: 0.3064 - acc: 0.8645 - val_loss: 0.5860 - val_acc: 0.7210\n",
      "Epoch 15/30\n",
      "100/100 [==============================] - 118s 1s/step - loss: 0.2852 - acc: 0.8805 - val_loss: 0.6275 - val_acc: 0.7190\n",
      "Epoch 16/30\n",
      "100/100 [==============================] - 123s 1s/step - loss: 0.2656 - acc: 0.8950 - val_loss: 0.6059 - val_acc: 0.7260\n",
      "Epoch 17/30\n",
      "100/100 [==============================] - 202s 2s/step - loss: 0.2369 - acc: 0.9115 - val_loss: 0.6493 - val_acc: 0.7200\n",
      "Epoch 18/30\n",
      "100/100 [==============================] - 219s 2s/step - loss: 0.2185 - acc: 0.9175 - val_loss: 0.6477 - val_acc: 0.7340\n",
      "Epoch 19/30\n",
      "100/100 [==============================] - 219s 2s/step - loss: 0.2006 - acc: 0.9230 - val_loss: 0.6419 - val_acc: 0.7310\n",
      "Epoch 20/30\n",
      "100/100 [==============================] - 216s 2s/step - loss: 0.1817 - acc: 0.9375 - val_loss: 0.7255 - val_acc: 0.7260\n",
      "Epoch 21/30\n",
      "100/100 [==============================] - 221s 2s/step - loss: 0.1595 - acc: 0.9420 - val_loss: 0.7021 - val_acc: 0.7240\n",
      "Epoch 22/30\n",
      "100/100 [==============================] - 223s 2s/step - loss: 0.1399 - acc: 0.9520 - val_loss: 0.8675 - val_acc: 0.7270\n",
      "Epoch 23/30\n",
      "100/100 [==============================] - 219s 2s/step - loss: 0.1242 - acc: 0.9560 - val_loss: 0.7459 - val_acc: 0.7380\n",
      "Epoch 24/30\n",
      "100/100 [==============================] - 210s 2s/step - loss: 0.1188 - acc: 0.9610 - val_loss: 0.8311 - val_acc: 0.7220\n",
      "Epoch 25/30\n",
      "100/100 [==============================] - 119s 1s/step - loss: 0.0977 - acc: 0.9655 - val_loss: 0.8335 - val_acc: 0.7260\n",
      "Epoch 26/30\n",
      "100/100 [==============================] - 116s 1s/step - loss: 0.0847 - acc: 0.9710 - val_loss: 0.9379 - val_acc: 0.7300\n",
      "Epoch 27/30\n",
      "100/100 [==============================] - 120s 1s/step - loss: 0.0713 - acc: 0.9800 - val_loss: 0.8462 - val_acc: 0.7400\n",
      "Epoch 28/30\n",
      "100/100 [==============================] - 119s 1s/step - loss: 0.0590 - acc: 0.9840 - val_loss: 0.9297 - val_acc: 0.7310\n",
      "Epoch 29/30\n",
      "100/100 [==============================] - 120s 1s/step - loss: 0.0539 - acc: 0.9855 - val_loss: 1.0150 - val_acc: 0.7200\n",
      "Epoch 30/30\n",
      "100/100 [==============================] - 119s 1s/step - loss: 0.0463 - acc: 0.9845 - val_loss: 1.0876 - val_acc: 0.7230\n"
     ]
    }
   ],
   "source": [
    "#利用批量生成器训练模型\n",
    "history = model.fit_generator( \n",
    "      train_generator, \n",
    "      steps_per_epoch=100, \n",
    "      epochs=30, \n",
    "      validation_data=validation_generator, \n",
    "      validation_steps=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#保存模型\n",
    "model.save('cats_and_dogs_small_1.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAEICAYAAACktLTqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f58840ead30>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAEICAYAAACktLTqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIpKFFNxFRLKQgruISBb6fzlzmMW6muWoAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f58808b4358>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 绘制训练过程中的损失曲线和精度曲线\n",
    "import matplotlib.pyplot as plt \n",
    "%matplotlib inline\n",
    "\n",
    "acc = history.history['acc'] \n",
    "val_acc = history.history['val_acc'] \n",
    "loss = history.history['loss'] \n",
    "val_loss = history.history['val_loss'] \n",
    "\n",
    "epochs = range(1, len(acc) + 1) \n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training acc') \n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy') \n",
    "plt.legend() \n",
    "\n",
    "plt.figure() \n",
    "\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss') \n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss') \n",
    "plt.legend() \n",
    "\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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}