是否存在全球平均池的反向功能？

Question

是否存在全局平均池的反向功能，我测试了上行，但不起作用

input_img = Input(shape=(nr_col, nr_rows, 1))
ec_conv1 = Conv2D(N[0], (3,3), activation='relu', padding='same')(input_img)
ec_pool1 = MaxPooling2D((2, 2), padding='same')(ec_conv1)
ec_conv2 = Conv2D(N[1], (3,3), activation='relu', padding='same')(ec_pool1)
ec_pool2 = MaxPooling2D((2, 2), padding='same')(ec_conv2)
ec_conv3 = Conv2D(N[2], (3,3), activation='relu', padding='same')(ec_pool2)
ec_pool3 = MaxPooling2D((2, 2), padding='same')(ec_conv3)
ec_conv4 = Conv2D(N[3], (3, 3), activation='relu', padding='same')(ec_pool3)
ec_ouput = MaxPooling2D((2, 2), padding='same')(ec_conv4)
z=GlobalAveragePooling2D(name='VC')(ec_ouput)
r=UpSampling2D(6,6)(z)