利用Keras识别TibetanMNIST数据集.ipynb

{"cells":[{"outputs":[{"output_type":"stream","text":"TibetanMNIST  TibetanMNIST.npz\tTibetanMNIST.tfrecords\r\n","name":"stdout"}],"execution_count":1,"source":"# 查看当前挂载的数据集目录\n!ls /home/kesci/input/TibetanMNIST5610/","cell_type":"code","metadata":{"trusted":true,"collapsed":false,"id":"547333A6D0CD4699BC27C6546BE83342","scrolled":false}},{"outputs":[{"output_type":"stream","text":"lost+found\r\n","name":"stdout"}],"execution_count":2,"source":"# 查看个人持久化工作区文件\n!ls /home/kesci/work/","cell_type":"code","metadata":{"trusted":true,"collapsed":false,"id":"C3E46D92B20240C5955155CEA44D9E7D","scrolled":false}},{"outputs":[],"execution_count":3,"source":"# 查看当前kernerl下的package\n# !pip list --format=columns","cell_type":"code","metadata":{"trusted":true,"collapsed":false,"id":"76C43959DBE0434BAE3F078B25F5BDB5","scrolled":false}},{"outputs":[],"execution_count":4,"source":"# 显示cell运行时长\n%load_ext klab-autotime","cell_type":"code","metadata":{"trusted":true,"collapsed":false,"id":"62A20ABFAC6F48A1BF9CEBAD0A0473F7","scrolled":false}},{"metadata":{"id":"89216D43349A42B69B1CE7E47ABD235F","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[{"output_type":"stream","text":"Using TensorFlow backend.\n","name":"stderr"},{"output_type":"stream","text":"time: 1.48 s\n","name":"stdout"}],"source":"import keras\nfrom keras.datasets import mnist\nfrom keras.models import Sequential\nfrom keras.layers import Dense, Dropout, Flatten\nfrom keras.layers import Conv2D, MaxPooling2D\nfrom keras import backend as K\nfrom keras.models import load_model\nimport pandas as pd\nimport numpy as np","execution_count":5},{"metadata":{"id":"E366224504B3406D88FC265005852A6C","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[{"output_type":"stream","text":"image： (17768, 28, 28)\nlabel： (17768,)\ntime: 108 ms\n","name":"stdout"}],"source":"img_rows, img_cols = 28, 28\n\ndata = np.load('/home/kesci/input/TibetanMNIST5610/TibetanMNIST.npz')\n\nimage_data, label_data = data['image'], data['label']\n\nprint('image：', image_data.shape)\nprint('label：', label_data.shape)\n","execution_count":6},{"metadata":{"id":"1ED606252E5448CB84BEF01997586DE8","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[{"output_type":"stream","text":"time: 391 ms\n","name":"stdout"}],"source":"from sklearn.model_selection import train_test_split\nX_train, X_test, Y_train, Y_test = train_test_split(image_data, label_data, test_size=0.2, random_state=42)","execution_count":7},{"metadata":{"id":"480CE8380AF348A3ACF457EA1C25FC20","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[{"output_type":"stream","text":"train_size: [1618, 1686, 1513, 1448, 1579, 1058, 1295, 1059, 1276, 1682]\ntrain_size: [378, 448, 359, 368, 388, 292, 323, 243, 366, 389]\ntime: 5.85 ms\n","name":"stdout"}],"source":"train_count = [0] * 10\ntest_count = [0] * 10\nfor tr_l in Y_train:\n    train_count[tr_l] += 1\nfor te_l in Y_test:\n    test_count[te_l] += 1\n    \nprint('train_size: {}'.format(train_count))\nprint('train_size: {}'.format(test_count))","execution_count":8},{"metadata":{"id":"756B8ABA820245DB8F13378C968C63E1","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[{"output_type":"stream","text":"time: 819 µs\n","name":"stdout"}],"source":"# Reshape image , Standardize , One-hot labels\nimg_rows, img_cols = 28, 28\ninput_shape = (img_rows, img_cols, 1) #tensorflow channels_last\nnum_classes = 10","execution_count":9},{"metadata":{"id":"C2194B620F694C76A948C5D072A7BA71","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[{"output_type":"stream","text":"time: 44.7 ms\n","name":"stdout"}],"source":"X_train = X_train.reshape(X_train.shape[0], img_rows, img_cols, 1).astype('float32')/255\nY_train = keras.utils.to_categorical(Y_train, num_classes)","execution_count":10},{"metadata":{"id":"BE7B0202E0B64A0A8D59AEDB350B9ED9","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[{"output_type":"stream","text":"time: 151 ms\n","name":"stdout"}],"source":"# Train model\nmodel = Sequential()\nmodel.add(Conv2D(32, kernel_size=(3, 3), activation='relu', input_shape=input_shape))\nmodel.add(Conv2D(64, (3, 3), activation='relu'))\nmodel.add(MaxPooling2D(pool_size=(2, 2)))\nmodel.add(Dropout(0.5))\nmodel.add(Flatten())\nmodel.add(Dense(128, activation='relu'))\nmodel.add(Dropout(0.5))\nmodel.add(Dense(num_classes, activation='softmax'))\n\n\n\nmodel.compile(loss=keras.losses.categorical_crossentropy, \n              optimizer=keras.optimizers.Adadelta(), \n              metrics=['accuracy'])","execution_count":11},{"metadata":{"id":"5262A373A2AE49C28BD225F0EF57C01C","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[{"output_type":"stream","text":"_________________________________________________________________\nLayer (type)                 Output Shape              Param #   \n=================================================================\nconv2d_1 (Conv2D)            (None, 26, 26, 32)        320       \n_________________________________________________________________\nconv2d_2 (Conv2D)            (None, 24, 24, 64)        18496     \n_________________________________________________________________\nmax_pooling2d_1 (MaxPooling2 (None, 12, 12, 64)        0         \n_________________________________________________________________\ndropout_1 (Dropout)          (None, 12, 12, 64)        0         \n_________________________________________________________________\nflatten_1 (Flatten)          (None, 9216)              0         \n_________________________________________________________________\ndense_1 (Dense)              (None, 128)               1179776   \n_________________________________________________________________\ndropout_2 (Dropout)          (None, 128)               0         \n_________________________________________________________________\ndense_2 (Dense)              (None, 10)                1290      \n=================================================================\nTotal params: 1,199,882\nTrainable params: 1,199,882\nNon-trainable params: 0\n_________________________________________________________________\ntime: 4.28 ms\n","name":"stdout"}],"source":"model.summary()","execution_count":12},{"metadata":{"id":"2E5B822DA7E04650A7B00C681CC56FDD","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[{"output_type":"stream","text":"time: 48.6 ms\n","name":"stdout"}],"source":"X_test = X_test.reshape(X_test.shape[0], img_rows, img_cols, 1).astype('float32')/255\nY_test_onehot = keras.utils.to_categorical(Y_test, num_classes)","execution_count":13},{"metadata":{"id":"76C47C504BA64B159FB118ADEF61EA68","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[{"output_type":"stream","text":"Train on 14214 samples, validate on 3554 samples\nEpoch 1/25\n14214/14214 [==============================] - 38s 3ms/step - loss: 0.6187 - acc: 0.7906 - val_loss: 0.3347 - val_acc: 0.8858\nEpoch 2/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.2335 - acc: 0.9229 - val_loss: 0.1310 - val_acc: 0.9530\nEpoch 3/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.1592 - acc: 0.9471 - val_loss: 0.0868 - val_acc: 0.9699\nEpoch 4/25\n14214/14214 [==============================] - 38s 3ms/step - loss: 0.1232 - acc: 0.9568 - val_loss: 0.1509 - val_acc: 0.9477\nEpoch 5/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.1042 - acc: 0.9660 - val_loss: 0.0749 - val_acc: 0.9733\nEpoch 6/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.0886 - acc: 0.9709 - val_loss: 0.0587 - val_acc: 0.9814\nEpoch 7/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.0810 - acc: 0.9719 - val_loss: 0.0678 - val_acc: 0.9783\nEpoch 8/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.0699 - acc: 0.9771 - val_loss: 0.0698 - val_acc: 0.9750\nEpoch 9/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.0624 - acc: 0.9785 - val_loss: 0.0534 - val_acc: 0.9823\nEpoch 10/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.0575 - acc: 0.9807 - val_loss: 0.0911 - val_acc: 0.9674\nEpoch 11/25\n14214/14214 [==============================] - 38s 3ms/step - loss: 0.0524 - acc: 0.9827 - val_loss: 0.0575 - val_acc: 0.9823\nEpoch 12/25\n14214/14214 [==============================] - 38s 3ms/step - loss: 0.0491 - acc: 0.9837 - val_loss: 0.0481 - val_acc: 0.9840\nEpoch 13/25\n14214/14214 [==============================] - 38s 3ms/step - loss: 0.0423 - acc: 0.9864 - val_loss: 0.0585 - val_acc: 0.9823\nEpoch 14/25\n14214/14214 [==============================] - 38s 3ms/step - loss: 0.0409 - acc: 0.9861 - val_loss: 0.0482 - val_acc: 0.9840\nEpoch 15/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.0411 - acc: 0.9866 - val_loss: 0.0474 - val_acc: 0.9842\nEpoch 16/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.0358 - acc: 0.9885 - val_loss: 0.0498 - val_acc: 0.9845\nEpoch 17/25\n14214/14214 [==============================] - 38s 3ms/step - loss: 0.0374 - acc: 0.9875 - val_loss: 0.0515 - val_acc: 0.9851\nEpoch 18/25\n14214/14214 [==============================] - 38s 3ms/step - loss: 0.0327 - acc: 0.9892 - val_loss: 0.0497 - val_acc: 0.9837\nEpoch 19/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.0324 - acc: 0.9892 - val_loss: 0.0475 - val_acc: 0.9868\nEpoch 20/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.0315 - acc: 0.9902 - val_loss: 0.0522 - val_acc: 0.9831\nEpoch 21/25\n14214/14214 [==============================] - 38s 3ms/step - loss: 0.0308 - acc: 0.9897 - val_loss: 0.0459 - val_acc: 0.9859\nEpoch 22/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.0291 - acc: 0.9898 - val_loss: 0.0453 - val_acc: 0.9885\nEpoch 23/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.0269 - acc: 0.9913 - val_loss: 0.0537 - val_acc: 0.9840\nEpoch 24/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.0293 - acc: 0.9913 - val_loss: 0.0539 - val_acc: 0.9854\nEpoch 25/25\n14214/14214 [==============================] - 37s 3ms/step - loss: 0.0251 - acc: 0.9916 - val_loss: 0.0546 - val_acc: 0.9854\ntime: 15min 36s\n","name":"stdout"}],"source":"history = model.fit(X_train, Y_train, \n                    batch_size=128, epochs=25, \n                    verbose=1, \n                    validation_data=(X_test, Y_test_onehot))","execution_count":14},{"metadata":{"id":"C3803A5DB78343318230DE46AD41A0B2","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[{"output_type":"stream","text":"time: 2.95 s\n","name":"stdout"}],"source":"Y_predict = model.predict_classes(X_test)","execution_count":15},{"metadata":{"id":"1099740371E141628C39103DB165014E","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[{"output_type":"execute_result","metadata":{},"data":{"text/plain":"0.9853685987619584"}},{"output_type":"stream","text":"time: 9.64 ms\n","name":"stdout"}],"source":"np.sum(Y_predict == Y_test)/len(Y_test)","execution_count":16},{"metadata":{"id":"281BC67013F24094A041D32323993834","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[{"output_type":"stream","text":"(3554, 28, 28, 1)\n(3554,)\n(3554,)\ntime: 704 µs\n","name":"stdout"}],"source":"print(X_test.shape)\nprint(Y_test.shape)\nprint(Y_predict.shape)","execution_count":17},{"metadata":{"id":"C5C1488158164DB39912FD085E5ED0B5","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[{"output_type":"stream","text":"time: 123 ms\n","name":"stdout"}],"source":"import matplotlib.pyplot as plt","execution_count":18},{"metadata":{"id":"1D62DB2A0FE44A0C8AAF8E3CEC062845","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[{"output_type":"display_data","metadata":{"needs_background":"light"},"data":{"text/plain":"<Figure size 648x576 with 9 Axes>","text/html":"<img src=\"https://cdn.kesci.com/rt_upload/1D62DB2A0FE44A0C8AAF8E3CEC062845/pkn7116zfu.png\">"}},{"output_type":"stream","text":"time: 579 ms\n","name":"stdout"}],"source":"figure = plt.figure(figsize=(9, 8))\nfor i in range(9):\n    plt.subplot(3, 3, i+1)\n    plt.imshow(X_test[i,:,:].reshape(28,28))\n    plt.title('识别结果是:{}, 实际是:{}'.format(Y_predict[i], Y_test[i]))\n    plt.axis('off')\nplt.show()","execution_count":19},{"metadata":{"id":"8E4093007DA94BC48AE8EE4214C95AD4","collapsed":false,"scrolled":false},"cell_type":"code","outputs":[],"source":"","execution_count":null}],"metadata":{"kernelspec":{"name":"python3","display_name":"Python 3","language":"python"},"language_info":{"name":"python","version":"3.6.4","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat":4,"nbformat_minor":0}