准确率99.3%——Keras进行藏文手写数字识别.ipynb

{"cells":[{"cell_type":"code","execution_count":1,"metadata":{},"outputs":[{"name":"stdout","output_type":"stream","text":["image数据： (17768, 28, 28)\n","Label数据： (17768,)\n"]},{"name":"stderr","output_type":"stream","text":["Using TensorFlow backend.\n"]}],"source":["from tensorflow import keras\n","from keras import utils\n","import numpy as np\n","\n","img_rows,img_cols = 28,28\n","\n","data = np.load('TibetanMNIST.npz')\n","\n","image_data, label_data= data['image'], data['label']\n","\n","print('image数据：',image_data.shape)\n","print('Label数据：',label_data.shape)"]},{"cell_type":"code","execution_count":8,"metadata":{},"outputs":[{"data":{"text/plain":["<matplotlib.image.AxesImage at 0x20e8b9ed400>"]},"execution_count":8,"metadata":{},"output_type":"execute_result"},{"data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{},"output_type":"display_data"}],"source":["import matplotlib.pyplot as plt\n","%matplotlib inline\n","\n","plt.imshow(image_data[0])"]},{"cell_type":"markdown","metadata":{},"source":["训练集处理并分割验证集"]},{"cell_type":"code","execution_count":3,"metadata":{},"outputs":[],"source":["x_train = image_data.reshape(17768, 28, 28, 1) / 255\n","y_train = utils.to_categorical(label_data, 10)\n","\n","from sklearn.model_selection import train_test_split\n","X_train,X_val,y_train,y_val = train_test_split(x_train,y_train,test_size=0.2,random_state=2018)"]},{"cell_type":"markdown","metadata":{},"source":["图片增强"]},{"cell_type":"code","execution_count":4,"metadata":{},"outputs":[],"source":["from keras.preprocessing.image import ImageDataGenerator\n","\n","train_gen = ImageDataGenerator(width_shift_range = 0.1, \n","                             height_shift_range = 0.1, \n","                             zoom_range = 0.2)\n","val_gen = ImageDataGenerator()\n","\n","batch_size = 64\n","train_generator = train_gen.flow(X_train,y_train,batch_size=batch_size, shuffle=False)\n","val_generator = val_gen.flow(X_val,y_val,batch_size=batch_size, shuffle=False)"]},{"cell_type":"markdown","metadata":{},"source":["创建模型"]},{"cell_type":"code","execution_count":5,"metadata":{},"outputs":[{"name":"stdout","output_type":"stream","text":["_________________________________________________________________\n","Layer (type)                 Output Shape              Param #   \n","=================================================================\n","input_1 (InputLayer)         (None, 28, 28, 1)         0         \n","_________________________________________________________________\n","conv2d_1 (Conv2D)            (None, 28, 28, 64)        640       \n","_________________________________________________________________\n","batch_normalization_1 (Batch (None, 28, 28, 64)        256       \n","_________________________________________________________________\n","conv2d_2 (Conv2D)            (None, 28, 28, 64)        36928     \n","_________________________________________________________________\n","batch_normalization_2 (Batch (None, 28, 28, 64)        256       \n","_________________________________________________________________\n","max_pooling2d_1 (MaxPooling2 (None, 14, 14, 64)        0         \n","_________________________________________________________________\n","conv2d_3 (Conv2D)            (None, 14, 14, 128)       73856     \n","_________________________________________________________________\n","batch_normalization_3 (Batch (None, 14, 14, 128)       512       \n","_________________________________________________________________\n","conv2d_4 (Conv2D)            (None, 14, 14, 128)       147584    \n","_________________________________________________________________\n","batch_normalization_4 (Batch (None, 14, 14, 128)       512       \n","_________________________________________________________________\n","max_pooling2d_2 (MaxPooling2 (None, 7, 7, 128)         0         \n","_________________________________________________________________\n","conv2d_5 (Conv2D)            (None, 7, 7, 256)         295168    \n","_________________________________________________________________\n","batch_normalization_5 (Batch (None, 7, 7, 256)         1024      \n","_________________________________________________________________\n","conv2d_6 (Conv2D)            (None, 7, 7, 256)         590080    \n","_________________________________________________________________\n","batch_normalization_6 (Batch (None, 7, 7, 256)         1024      \n","_________________________________________________________________\n","max_pooling2d_3 (MaxPooling2 (None, 3, 3, 256)         0         \n","_________________________________________________________________\n","flatten_1 (Flatten)          (None, 2304)              0         \n","_________________________________________________________________\n","dropout_1 (Dropout)          (None, 2304)              0         \n","_________________________________________________________________\n","dense_1 (Dense)              (None, 10)                23050     \n","_________________________________________________________________\n","activation_1 (Activation)    (None, 10)                0         \n","=================================================================\n","Total params: 1,170,890\n","Trainable params: 1,169,098\n","Non-trainable params: 1,792\n","_________________________________________________________________\n"]}],"source":["from keras.layers import *\n","from keras.models import *\n","from keras.callbacks import ModelCheckpoint,ReduceLROnPlateau\n","def build_model(out_dims, img_size):\n","    inputs_dim = Input((img_size, img_size, 1))\n","\n","    x = Conv2D(64, (3, 3), strides=(1, 1), padding='same', activation = 'relu')(inputs_dim)\n","    x = BatchNormalization()(x)\n","    x = Conv2D(64, (3, 3), strides=(1, 1), padding='same',activation = 'relu')(x)\n","    x = BatchNormalization()(x)\n","    x = MaxPool2D(pool_size=(2, 2))(x)\n","    \n","    x = Conv2D(128, (3, 3), strides=(1, 1), padding='same',activation = 'relu')(x)\n","    x = BatchNormalization()(x)\n","    x = Conv2D(128, (3, 3), strides=(1, 1), padding='same',activation = 'relu')(x)\n","    x = BatchNormalization()(x)\n","    x = MaxPool2D(pool_size=(2, 2))(x)\n","    \n","    x = Conv2D(256, (3, 3), strides=(1, 1), padding='same',activation = 'relu')(x)\n","    x = BatchNormalization()(x)\n","    x = Conv2D(256, (3, 3), strides=(1, 1), padding='same',activation = 'relu')(x)\n","    x = BatchNormalization()(x)\n","    x = MaxPool2D(pool_size=(2, 2))(x)\n","    \n","    x_flat = Flatten()(x)\n","    dp_1 = Dropout(0.5)(x_flat)\n","    fc2 = Dense(out_dims)(dp_1)\n","    fc2 = Activation('softmax')(fc2)\n","\n","    model = Model(inputs=inputs_dim, outputs=fc2)\n","    return model\n","\n","model = build_model(out_dims=10, img_size=28)\n","model.summary()"]},{"cell_type":"markdown","metadata":{},"source":["训练模型，保存最低val_loss时的模型权重，并设置在2个epoch没有loss下降的时候降低学习率学习率"]},{"cell_type":"code","execution_count":6,"metadata":{},"outputs":[{"name":"stdout","output_type":"stream","text":["Epoch 1/20\n","223/223 [==============================] - 16s 72ms/step - loss: 0.5862 - acc: 0.8526 - val_loss: 0.1630 - val_acc: 0.9586\n","\n","Epoch 00001: val_loss improved from inf to 0.16299, saving model to TibetanMNIST.hdf5\n","Epoch 2/20\n","223/223 [==============================] - 14s 62ms/step - loss: 0.2381 - acc: 0.9366 - val_loss: 0.1604 - val_acc: 0.9555\n","\n","Epoch 00002: val_loss improved from 0.16299 to 0.16040, saving model to TibetanMNIST.hdf5\n","Epoch 3/20\n","223/223 [==============================] - 14s 62ms/step - loss: 0.1936 - acc: 0.9487 - val_loss: 0.0574 - val_acc: 0.9806\n","\n","Epoch 00003: val_loss improved from 0.16040 to 0.05745, saving model to TibetanMNIST.hdf5\n","Epoch 4/20\n","223/223 [==============================] - 14s 62ms/step - loss: 0.1440 - acc: 0.9605 - val_loss: 0.2429 - val_acc: 0.9451\n","\n","Epoch 00004: val_loss did not improve from 0.05745\n","Epoch 5/20\n","223/223 [==============================] - 14s 62ms/step - loss: 0.1253 - acc: 0.9673 - val_loss: 0.0552 - val_acc: 0.9845\n","\n","Epoch 00005: val_loss improved from 0.05745 to 0.05521, saving model to TibetanMNIST.hdf5\n","Epoch 6/20\n","223/223 [==============================] - 14s 62ms/step - loss: 0.1103 - acc: 0.9694 - val_loss: 0.0998 - val_acc: 0.9733\n","\n","Epoch 00006: val_loss did not improve from 0.05521\n","Epoch 7/20\n","223/223 [==============================] - 14s 62ms/step - loss: 0.1001 - acc: 0.9725 - val_loss: 0.0718 - val_acc: 0.9820\n","\n","Epoch 00007: val_loss did not improve from 0.05521\n","\n","Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.00020000000949949026.\n","Epoch 8/20\n","223/223 [==============================] - 14s 63ms/step - loss: 0.0561 - acc: 0.9832 - val_loss: 0.0327 - val_acc: 0.9913\n","\n","Epoch 00008: val_loss improved from 0.05521 to 0.03265, saving model to TibetanMNIST.hdf5\n","Epoch 9/20\n","223/223 [==============================] - 15s 65ms/step - loss: 0.0464 - acc: 0.9873 - val_loss: 0.0297 - val_acc: 0.9913\n","\n","Epoch 00009: val_loss improved from 0.03265 to 0.02970, saving model to TibetanMNIST.hdf5\n","Epoch 10/20\n","223/223 [==============================] - 14s 64ms/step - loss: 0.0408 - acc: 0.9868 - val_loss: 0.0247 - val_acc: 0.9930\n","\n","Epoch 00010: val_loss improved from 0.02970 to 0.02467, saving model to TibetanMNIST.hdf5\n","Epoch 11/20\n","223/223 [==============================] - 14s 64ms/step - loss: 0.0376 - acc: 0.9887 - val_loss: 0.0391 - val_acc: 0.9890\n","\n","Epoch 00011: val_loss did not improve from 0.02467\n","Epoch 12/20\n","223/223 [==============================] - 14s 64ms/step - loss: 0.0329 - acc: 0.9903 - val_loss: 0.0280 - val_acc: 0.9927\n","\n","Epoch 00012: val_loss did not improve from 0.02467\n","\n","Epoch 00012: ReduceLROnPlateau reducing learning rate to 4.0000001899898055e-05.\n","Epoch 13/20\n","223/223 [==============================] - 14s 62ms/step - loss: 0.0302 - acc: 0.9902 - val_loss: 0.0270 - val_acc: 0.9921\n","\n","Epoch 00013: val_loss did not improve from 0.02467\n","Epoch 14/20\n","223/223 [==============================] - 14s 63ms/step - loss: 0.0293 - acc: 0.9915 - val_loss: 0.0282 - val_acc: 0.9921\n","\n","Epoch 00014: val_loss did not improve from 0.02467\n","\n","Epoch 00014: ReduceLROnPlateau reducing learning rate to 8.000000525498762e-06.\n","Epoch 15/20\n","223/223 [==============================] - 14s 62ms/step - loss: 0.0318 - acc: 0.9893 - val_loss: 0.0280 - val_acc: 0.9913\n","\n","Epoch 00015: val_loss did not improve from 0.02467\n","Epoch 16/20\n","223/223 [==============================] - 14s 63ms/step - loss: 0.0268 - acc: 0.9908 - val_loss: 0.0277 - val_acc: 0.9913\n","\n","Epoch 00016: val_loss did not improve from 0.02467\n","\n","Epoch 00016: ReduceLROnPlateau reducing learning rate to 1.6000001778593287e-06.\n","Epoch 17/20\n","223/223 [==============================] - 14s 62ms/step - loss: 0.0271 - acc: 0.9911 - val_loss: 0.0275 - val_acc: 0.9913\n","\n","Epoch 00017: val_loss did not improve from 0.02467\n","Epoch 18/20\n","223/223 [==============================] - 14s 63ms/step - loss: 0.0287 - acc: 0.9907 - val_loss: 0.0276 - val_acc: 0.9913\n","\n","Epoch 00018: val_loss did not improve from 0.02467\n","\n","Epoch 00018: ReduceLROnPlateau reducing learning rate to 3.200000264769187e-07.\n","Epoch 19/20\n","223/223 [==============================] - 14s 63ms/step - loss: 0.0256 - acc: 0.9921 - val_loss: 0.0277 - val_acc: 0.9913\n","\n","Epoch 00019: val_loss did not improve from 0.02467\n","Epoch 20/20\n","223/223 [==============================] - 14s 61ms/step - loss: 0.0262 - acc: 0.9924 - val_loss: 0.0277 - val_acc: 0.9907\n","\n","Epoch 00020: val_loss did not improve from 0.02467\n","\n","Epoch 00020: ReduceLROnPlateau reducing learning rate to 6.400000529538374e-08.\n"]}],"source":["checkpointer = ModelCheckpoint(filepath='TibetanMNIST.hdf5', \n","                              verbose=1, save_best_only=True)\n","reduce = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=2,verbose=1,mode='min')\n","model.compile(optimizer='adam',\n","              loss='categorical_crossentropy',\n","              metrics=['accuracy'])\n","\n","epochs = 20\n","\n","history = model.fit_generator(train_generator,\n","          validation_data=val_generator,\n","          epochs=epochs,callbacks=[checkpointer,reduce],verbose=1)"]},{"cell_type":"code","execution_count":9,"metadata":{},"outputs":[{"data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAYsAAAEWCAYAAACXGLsWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngg==\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{},"output_type":"display_data"},{"data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAYUAAAEWCAYAAACJ0YulAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{},"output_type":"display_data"}],"source":["plt.plot(history.history['acc'])\n","plt.plot(history.history['val_acc'])\n","plt.title('model accuracy')\n","plt.ylabel('accuracy')\n","plt.xlabel('epoch')\n","plt.legend(['train', 'val'], loc='upper left')\n","plt.show()\n","# summarize history for loss\n","plt.plot(history.history['loss'])\n","plt.plot(history.history['val_loss'])\n","plt.title('model loss')\n","plt.ylabel('loss')\n","plt.xlabel('epoch')\n","plt.legend(['train', 'val'], loc='upper left')\n","plt.show()"]},{"cell_type":"markdown","metadata":{},"source":["在第10个epoch达到val_acc: 0.9930，准确率99.3%"]},{"cell_type":"code","execution_count":null,"metadata":{},"outputs":[],"source":[]}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python","nbconvert_exporter":"python","file_extension":".py","version":"3.5.2","pygments_lexer":"ipython3"}},"nbformat":4,"nbformat_minor":2}