tensorflow2/kera搭建网络模型报错Input layers to a `Model` must be `InputLayer` objects. Received inputs:

报错内容

Input layers to a `Model` must be `InputLayer` objects. 
Received inputs: Tensor("batch_normalization_1/cond/Merge:0", 
shape=(?, 321, 64), dtype=float32). 
Input 0 (0-based) originates from layer type `BatchNormalization`.

原程序

def build_disk_and_q_net(img_shape):
    img = Input(shape=img_shape)

    # Shared layers between discriminator and recognition network
    img = Conv1D(64, 10, strides=2)(img)
    # img = Conv1D(64, 10, strides=2, input_shape=img_shape)
    img = BatchNormalization(momentum=0.8)(img)
    layer1 = LeakyReLU(alpha=0.2)(img)

修改方法：因为输入是img， 所以网络连接层换一个变量名（比如x）就好了

def build_disk_and_q_net(img_shape):
    img = Input(shape=img_shape)

    # Shared layers between discriminator and recognition network
    x = Conv1D(64, 10, strides=2)(img)
    # img = Conv1D(64, 10, strides=2, input_shape=img_shape)
    x = BatchNormalization(momentum=0.8)(x)
    layer1 = LeakyReLU(alpha=0.2)(x)