class C(B): def method(self, arg): super(C, self).method(arg)
子类C重写了父类B中同名方法method,在重写的实现中通过super实例化的代理对象调用父类的同名方法。
举例如下
class BaseSuperResolutionModel(object):
def __init__(self, model_name, scale_factor):
"""
Base model to provide a standard interface of adding Super Resolution models
"""
self.model = None # type: Model
self.model_name = model_name
self.scale_factor = scale_factor
self.weight_path = None
self.type_scale_type = "norm" # Default = "norm" = 1. / 255
self.type_requires_divisible_shape = False
self.type_true_upscaling = False
self.evaluation_func = None
self.uses_learning_phase = False
def create_model(self, height=32, width=32, channels=3, load_weights=False, batch_size=128) -> Model:
"""
Subclass dependent implementation.
"""
if self.type_requires_divisible_shape:
assert height * img_utils._image_scale_multiplier % 4 == 0, "Height of the image must be divisible by 4"
assert width * img_utils._image_scale_multiplier % 4 == 0, "Width of the image must be divisible by 4"
if K.image_dim_ordering() == "th":
shape = (channels, width * img_utils._image_scale_multiplier, height * img_utils._image_scale_multiplier)
else:
shape = (width * img_utils._image_scale_multiplier, height * img_utils._image_scale_multiplier, channels)
init = Input(shape=shape)
return init
class ImageSuperResolutionModel(BaseSuperResolutionModel):
def __init__(self, scale_factor):
super(ImageSuperResolutionModel, self).__init__("Image SR", scale_factor)
self.f1 = 9
self.f2 = 1
self.f3 = 5
self.n1 = 64
self.n2 = 32
self.weight_path = "weights/SR Weights %dX.h5" % (self.scale_factor)
def create_model(self, height=32, width=32, channels=3, load_weights=False, batch_size=128):
"""
Creates a model to be used to scale images of specific height and width.
"""
init = super(ImageSuperResolutionModel, self).create_model(height, width, channels, load_weights, batch_size)
x = Convolution2D(self.n1, self.f1, self.f1, activation='relu', border_mode='same', name='level1')(init)
x = Convolution2D(self.n2, self.f2, self.f2, activation='relu', border_mode='same', name='level2')(x)
out = Convolution2D(channels, self.f3, self.f3, border_mode='same', name='output')(x)
model = Model(init, out)
adam = optimizers.Adam(lr=1e-3)
model.compile(optimizer=adam, loss='mse', metrics=[PSNRLoss])
if load_weights: model.load_weights(self.weight_path)
self.model = model
return model
super(ImageSuperResolutionModel, self).__init__("Image SR", scale_factor) 继承了基类中的初始化方法