手写识别训练

最新推荐文章于 2023-06-25 08:35:47 发布

chilian12321

最新推荐文章于 2023-06-25 08:35:47 发布

阅读量280

点赞数

本文链接：https://blog.csdn.net/chilian12321/article/details/108062873

版权

python 同时被 2 个专栏收录

3 篇文章 1 订阅

订阅专栏

deep learning

1 篇文章 0 订阅

订阅专栏

手写识别--使用keras框架，构建CNN网络完成对MNIST数据集的训练，评估及预测

#cnn网络 --mnist
import tensorflow as tf
import  keras
from  keras.datasets import mnist
from  keras import backend as K
from keras.models import Sequential
from keras.layers import Dense, Dropout, Flatten
from keras.layers import Conv2D, MaxPooling2D

batch_size=128
num_classes = 10
epochs = 5
img_rows, img_cols = 28, 28 #输入图像的大小
(x_train, y_train), (x_test, y_test) = mnist.load_data()#数据集洗牌

#1.1整理数据
if K.image_data_format() == 'channels_first':
    x_train = x_train.reshape(x_train.shape[0], 1, img_rows, img_cols)
    x_test = x_test.reshape(x_test.shape[0], 1, img_rows, img_cols)
    input_shape = (1, img_rows, img_cols)
else:
    x_train = x_train.reshape(x_train.shape[0], img_rows, img_cols, 1)
    x_test = x_test.reshape(x_test.shape[0], img_rows, img_cols, 1)
    input_shape = (img_rows, img_cols, 1)
x_train = x_train.astype('float32')
x_test = x_test.astype('float32')
# 幅度缩放
x_train /= 255
x_test /= 255
print('x_train shape:', x_train.shape)
print(x_train.shape[0], 'train samples')
print(x_test.shape[0], 'test samples')
# convert class vectors to binary class matrices
# 对数据进行转换，因为在keras中训练等对数据的数量一般要用numpy.ndarray类型
y_train = keras.utils.to_categorical(y_train, num_classes)
y_test =  keras.utils.to_categorical(y_test, num_classes)
# 查看一下数据
import matplotlib.pyplot as plt

#%matplotlib inline
plt.imshow(x_train[0].reshape(28, 28))


## 1.2 模型部分 创建一个模型
model=Sequential()
#卷积层 3*3的卷积核，后面接激活函数，rule.32个神经元
model.add(Conv2D(32, kernel_size=(3, 3),
                 activation='relu',
                 input_shape=input_shape))
model.add(Conv2D(64, (3, 3), activation='relu'))
#池化层
model.add(MaxPooling2D(pool_size=(2, 2)))
#正则化
model.add(Dropout(0.25))
# 全连接层前面的shape要是n*1的，就是类似向量形式，所以沿展开
model.add(Flatten())
# 全连接
model.add(Dense(128, activation='relu'))
model.add(Dropout(0.5))
# 全连接，直接连接到多少个类
model.add(Dense(num_classes, activation='softmax'))

## 1.3 编译
model.compile(loss=keras.losses.categorical_crossentropy,
              optimizer=keras.optimizers.Adadelta(),
              metrics=['accuracy'])
## 1.4 训练
model.fit(x_train, y_train,
          batch_size=batch_size,
          epochs=epochs,
          verbose=1,
          validation_data=(x_test, y_test))
## 1.5 评估


score = model.evaluate(x_test, y_test, verbose=0)
print('Test loss:', score[0])
print('Test accuracy:', score[1])