Tensorflow：不支持的操作数类型 -：'Sequential' 和 'Sequential'

Question

def build_net(img_shape):

    """

    :type img_shape: tuple. Shape of input image. Here is(1,height, width). 1 because pgm file only has one channel.

    :rtype:tensorflow Sequential

    """

    model = tf.keras.Sequential()

    # convolution layer 1

    model.add(tf.keras.layers.Conv2D(filters = 16, kernel_size = 3, strides = 1, activation = "relu", input_shape = img_shape, data_format = "channels_first"))

    model.add(tf.keras.layers.MaxPool2D(pool_size = 2))

    model.add(tf.keras.layers.Dropout(0.1))

    # convolution layer 2

    model.add(tf.keras.layers.Conv2D(filters = 32, kernel_size = 3, strides = 1))

    model.add(tf.keras.layers.MaxPool2D(pool_size = 2))

    model.add(tf.keras.layers.Dropout(0.1))

    model.add(tf.keras.layers.Flatten())

    model.add(tf.keras.layers.Dense(1024))

    model.add(tf.keras.layers.Dropout(0.25))

    model.add(tf.keras.layers.Dense(512, activation='relu'))

    # deep face mentioned that there are 67 points to detect on a human face, so use 70 features.

    model.add(tf.keras.layers.Dense(70, activation='relu'))

    print(model.summary())

    return model

并定义 adist来计算两个输出向量之间的距离。

im1_features = build_net(input_dim)

im2_features = build_net(input_dim)

dist = tf.keras.layers.Lambda(lambda tensors: tf.keras.backend.abs[tensors[0] - tensors[1]])([im1_features, im2_features])

错误发生在dist

  File "e:\School\AIAS\proj\build_model.py", line 102, in <lambda>

    dist = tf.keras.layers.Lambda(lambda tensors: tf.keras.backend.abs[tensors[0] - tensors[1]])([im1_features, im2_features])

TypeError: unsupported operand type(s) for -: 'Sequential' and 'Sequential'

如何使函数build_net返回向量而不是 Sequential 对象？

Answer 1

尝试tf.keras.abs这样调用：tf.keras.backend.abs( &nbsp;&nbsp;&nbsp;&nbsp;x )不是tf.keras.backend.abs[ &nbsp;&nbsp;&nbsp;&nbsp;x ]它是一个函数，而不是一个数组。这是否解决了您的问题？