tf 常用函数

中文教程：https://tensorflow.google.cn/api_docs/python/tf 

     tf.keras.layers： Input, Conv2D, BatchNormalization, ReLU, Add, add, Concatenate, Lambda, Dense, Flatten, MaxPooling2D,  AveragePooling2D, ZeroPadding2D, Activation,  DepthwiseConv2D, Dropout,  GlobalAveragePooling2D,  GlobalMaxPool2D, Dot, Conv2DTranspose，

results = tf.keras.layers.Lambda(lambda x: tf.image.resize(x[0], (tf.shape(x[1])[1], tf.shape(x[1])[2]), method = tf.image.ResizeMethod.BILINEAR))([results, inputs])

tf.keras.layers.Activation('relu')(x)

tf.keras.layers.Conv2D(filters2, kernel_size, padding='same', dilation_rate = dilation_rate, name=conv_name_base + '2b', use_bias=False)(x)

tf.keras.layers.Conv2D(n, kernel_size = (3,3), padding = 'same', activation = 'relu', kernel_initializer