Update cnn_function.py

cnn_function.py

+import numpy as np
+'''
+def get_data():
+    """
+    数据预处理
+    """
+    # 获取原始数据
+    train_x_ori, train_y_set, test_x_ori, test_y_set, classes = utils.load_data_sets()
+    # m_train: 训练集样本数量
+    m_train = train_x_ori.shape[0]
+    # m_test: 测试集样本数量
+    m_test = test_x_ori.shape[0]
+    # 图片样本长宽像素数量
+    num_px_x = train_x_ori.shape[1]
+    num_px_y = train_x_ori.shape[2]
 
+    # 定义输入数据维度，注意样本图像是3通道
+    DATA_DIM = num_px_x * num_px_y * 3
+
+    # 转换数据形状为
+    train_x_flatten = train_x_ori.reshape(m_train, -1)
+    test_x_flatten = test_x_ori.reshape(m_test, -1)
+
+    # 归一化处理
+    train_x_set = train_x_flatten / 255
+    test_x_set = test_x_flatten / 255
+
+    # 合并数据
+    train_set = np.hstack((train_x_set, train_y_set.T))
+    test_set = np.hstack((test_x_set, test_y_set.T))
+
+    return train_set, test_set, DATA_DIM, classes
+'''
+data1=[1,2,3,4,5,6,7,8,9]
+
+def read_data(data_set):
+    """
+    构造reader
+    迭代器
+    :param data_set: 要获取的数据的数据集
+    :return: reader: 用户返回训练数据及数据标签的生成器(generator)
+    """
+    def reader():
+        """
+        一个reader生成器
+        :return: 每次训练数据及数据标签
+        data[:-1]: 训练数据
+        data[-1:]: 数据标签
+        """
+        for data in data_set:
+            yield data[:-1],data[-1:]
+    return reader
+#reader=read_data(data1)
+#reader1=reader()
+#print(type(reader1))