Question

我想创建一个自定义层来计算Keras中的梯度。我写了定制层，但是失败了。该概念之后是grad-cam，我打算使用渐变图进行上采样。

代码如下。

from keras import backend as K
from keras.engine.topology import Layer

class grad_cam(Layer):

    def __init__(self, output_dim, **kwargs):
        self.output_dim = output_dim
        super(grad_cam, self).__init__(**kwargs)

    def build(self, input_shape):
        assert isinstance(input_shape, list)
        
        self.kernel = self.add_weight(name='kernel',
                                      shape=(input_shape[0][1], self.output_dim),
                                      initializer='uniform',
                                      trainable=True)
        super(grad_cam, self).build(input_shape)  

    def call(self, x):
        assert isinstance(x, list)
        last_layer, conv_layer = x
        loss                 = K.max(last_layer)
        grad_wrt_fmap        = K.gradients(loss,conv_layer)
        return K.dot(grad_wrt_fmap, self.kernel)

    def compute_output_shape(self, input_shape):
        assert isinstance(input_shape, list)
        shape_a, shape_b = input_shape
        return shape_b

模型架构：

base_model = DenseNet121(include_top = False, weights='imagenet')
# add a global spatial average pooling layer
x = base_model.output
x = GlobalAveragePooling2D()(x)
# let's add a fully-connected layer
x = Dense(1024, activation='relu')(x)
# and a logistic layer
predictions = Dense(len(sequence), activation='sigmoid')(x)
predictions = grad_cam(base_model.output.shape)([predictions, base_model.output])
print(predictions)
model = Model(inputs=base_model.input, outputs=predictions)
model.summary()

但是，我遇到了错误问题。

ValueError: Can't convert Python sequence with mixed types to Tensor.

我不知道如何解决它。请帮我解决！谢谢！

在Keras中建立一个自定义的计算梯度层

0 个答案: