实战keras——用CNN实现cifar10图像分类
Posted ustc-zcc
tags:
篇首语：本文由小常识网(cha138.com)小编为大家整理，主要介绍了实战keras——用CNN实现cifar10图像分类相关的知识，希望对你有一定的参考价值。
原文：https://blog.csdn.net/zzulp/article/details/76358694
import keras from keras.datasets import cifar10 from keras.models import Sequential from keras.layers import Dense, Dropout, Activation, Flatten from keras.layers import Conv2D, MaxPooling2D num_classes = 10 model_name = ‘cifar10.h5‘ # The data, shuffled and split between train and test sets: (x_train, y_train), (x_test, y_test) = cifar10.load_data() x_train = x_train.astype(‘float32‘)/255 x_test = x_test.astype(‘float32‘)/255 # Convert class vectors to binary class matrices. y_train = keras.utils.to_categorical(y_train, num_classes) y_test = keras.utils.to_categorical(y_test, num_classes) model = Sequential() model.add(Conv2D(32, (3, 3), padding=‘same‘, input_shape=x_train.shape[1:])) model.add(Activation(‘relu‘)) model.add(MaxPooling2D(pool_size=(2, 2))) model.add(Dropout(0.25)) model.add(Conv2D(64, (3, 3), padding=‘same‘)) model.add(Activation(‘relu‘)) model.add(MaxPooling2D(pool_size=(2, 2))) model.add(Dropout(0.25)) model.add(Flatten()) model.add(Dense(512)) model.add(Activation(‘relu‘)) model.add(Dropout(0.5)) model.add(Dense(num_classes)) model.add(Activation(‘softmax‘)) model.summary() # initiate RMSprop optimizer opt = keras.optimizers.rmsprop(lr=0.001, decay=1e-6) # train the model using RMSprop model.compile(loss=‘categorical_crossentropy‘, optimizer=opt, metrics=[‘accuracy‘]) hist = model.fit(x_train, y_train, epochs=40, shuffle=True) model.save(model_name) # evaluate loss, accuracy = model.evaluate(x_test, y_test) print(loss, accuracy)
实验结果：
Downloading data from http://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz 170475520/170498071 [============================>.] - ETA: 0s_________________________________________________________________ Layer (type) Output Shape Param # ================================================================= conv2d_1 (Conv2D) (None, 32, 32, 32) 896 _________________________________________________________________ activation_1 (Activation) (None, 32, 32, 32) 0 _________________________________________________________________ max_pooling2d_1 (MaxPooling2 (None, 16, 16, 32) 0 _________________________________________________________________ dropout_1 (Dropout) (None, 16, 16, 32) 0 _________________________________________________________________ conv2d_2 (Conv2D) (None, 16, 16, 64) 18496 _________________________________________________________________ activation_2 (Activation) (None, 16, 16, 64) 0 _________________________________________________________________ max_pooling2d_2 (MaxPooling2 (None, 8, 8, 64) 0 _________________________________________________________________ dropout_2 (Dropout) (None, 8, 8, 64) 0 _________________________________________________________________ flatten_1 (Flatten) (None, 4096) 0 _________________________________________________________________ dense_1 (Dense) (None, 512) 2097664 _________________________________________________________________ activation_3 (Activation) (None, 512) 0 _________________________________________________________________ dropout_3 (Dropout) (None, 512) 0 _________________________________________________________________ dense_2 (Dense) (None, 10) 5130 _________________________________________________________________ activation_4 (Activation) (None, 10) 0 ================================================================= Total params: 2,122,186 Trainable params: 2,122,186 Non-trainable params: 0 _________________________________________________________________ Epoch 1/40 50000/50000 [==============================] - 189s - loss: 1.5264 - acc: 0.4558 Epoch 2/40 50000/50000 [==============================] - 185s - loss: 1.2152 - acc: 0.5769 Epoch 3/40 50000/50000 [==============================] - 192s - loss: 1.1367 - acc: 0.6118 Epoch 4/40 50000/50000 [==============================] - 183s - loss: 1.1145 - acc: 0.6241 Epoch 5/40 50000/50000 [==============================] - 189s - loss: 1.1131 - acc: 0.6273 Epoch 6/40 50000/50000 [==============================] - 192s - loss: 1.1175 - acc: 0.6313 Epoch 7/40 50000/50000 [==============================] - 202s - loss: 1.1309 - acc: 0.6299 Epoch 8/40 50000/50000 [==============================] - 187s - loss: 1.1406 - acc: 0.6278 Epoch 9/40 50000/50000 [==============================] - 190s - loss: 1.1583 - acc: 0.6221 Epoch 10/40 50000/50000 [==============================] - 188s - loss: 1.1689 - acc: 0.6199 Epoch 11/40 50000/50000 [==============================] - 183s - loss: 1.1896 - acc: 0.6134 Epoch 12/40 50000/50000 [==============================] - 188s - loss: 1.2032 - acc: 0.6101 Epoch 13/40 50000/50000 [==============================] - 186s - loss: 1.2246 - acc: 0.6011 Epoch 14/40 50000/50000 [==============================] - 192s - loss: 1.2405 - acc: 0.6000 Epoch 15/40 50000/50000 [==============================] - 170s - loss: 1.2514 - acc: 0.5958 Epoch 16/40 50000/50000 [==============================] - 172s - loss: 1.2627 - acc: 0.5912 Epoch 17/40 50000/50000 [==============================] - 177s - loss: 1.2835 - acc: 0.5838 Epoch 18/40 50000/50000 [==============================] - 179s - loss: 1.2876 - acc: 0.5809 Epoch 19/40 50000/50000 [==============================] - 180s - loss: 1.3085 - acc: 0.5782 Epoch 20/40 50000/50000 [==============================] - 180s - loss: 1.3253 - acc: 0.5695 Epoch 21/40 50000/50000 [==============================] - 180s - loss: 1.3375 - acc: 0.5651 Epoch 22/40 50000/50000 [==============================] - 183s - loss: 1.3483 - acc: 0.5623 Epoch 23/40 50000/50000 [==============================] - 177s - loss: 1.3567 - acc: 0.5599 Epoch 24/40 50000/50000 [==============================] - 178s - loss: 1.3697 - acc: 0.5541 Epoch 25/40 50000/50000 [==============================] - 178s - loss: 1.3722 - acc: 0.5518 Epoch 26/40 50000/50000 [==============================] - 181s - loss: 1.3848 - acc: 0.5479 Epoch 27/40 50000/50000 [==============================] - 181s - loss: 1.3916 - acc: 0.5474 Epoch 28/40 50000/50000 [==============================] - 183s - loss: 1.4081 - acc: 0.5403 Epoch 29/40 50000/50000 [==============================] - 172s - loss: 1.4229 - acc: 0.5387 Epoch 30/40 50000/50000 [==============================] - 190s - loss: 1.4153 - acc: 0.5383 Epoch 31/40 50000/50000 [==============================] - 183s - loss: 1.4355 - acc: 0.5324 Epoch 32/40 50000/50000 [==============================] - 191s - loss: 1.4667 - acc: 0.5251 Epoch 33/40 50000/50000 [==============================] - 169s - loss: 1.4690 - acc: 0.5188 Epoch 34/40 50000/50000 [==============================] - 168s - loss: 1.4798 - acc: 0.5176 Epoch 35/40 50000/50000 [==============================] - 181s - loss: 1.5152 - acc: 0.5054 Epoch 36/40 50000/50000 [==============================] - 173s - loss: 1.4985 - acc: 0.5067 Epoch 37/40 50000/50000 [==============================] - 182s - loss: 1.5030 - acc: 0.5098 Epoch 38/40 50000/50000 [==============================] - 178s - loss: 1.5298 - acc: 0.4967 Epoch 39/40 50000/50000 [==============================] - 181s - loss: 1.5237 - acc: 0.5014 Epoch 40/40 50000/50000 [==============================] - 181s - loss: 1.4933 - acc: 0.5103 9952/10000 [============================>.] - ETA: 0s1.80146283646 0.3274
以上是关于实战keras——用CNN实现cifar10图像分类的主要内容，如果未能解决你的问题，请参考以下文章
Keras CIFAR-10分类 自定义simple CNN篇
小白学习keras教程六基于CIFAR-10数据集训练CNN-RNN神经网络模型
Keras 教程示例 cifar10_cnn.py 在 windows 10 上的 200 个 epoch 中给出 acc: 0.34 val_acc: 0.40？