core

Classes

Unet

Functions

build_Unet(input_shape[, n_c_fltr, k_size, p_size, ...])

Build the Unet.

Module Contents

core.build_Unet(input_shape, n_c_fltr=128, k_size=(3, 3), p_size=(2, 2), act_reg=None, act_reg_factor=0.0001)

Build the Unet.

Construct the architecture of the Unet model based on the given parameters.

Parameters:

  • input_shape (tuple of int) – The shape of the input (n_lat, n_lon, n_in_vars), where n_lat is the number of latitude grid cells, n_lon is the number of longitude grid cells, and n_in_vars is the number of input variables.

  • n_c_fltr (int, optional) – The number of filters for the initial convolutional layer.

  • k_size (tuple of int, optional) – The kernel size to use in the convolutional layers.

  • p_size (tuple of int, optional) – The factors by which to downscale in the Max Pooling layers.

  • act_reg (str, optional) – The name of the regularization function (of type keras.regularizers.Regularizer) to apply to the activations of the layers.

  • act_reg_factor (float, optional) – The regularization factor to apply to the activations of the layers.

Returns:

model – The constructed Unet model.

Return type:

keras.Model

class core.Unet
model = None
build(input_shape, **kwargs)
compile(optimizer, loss, **kwargs)
info()
train(*args, **kwargs)
predict(x)
summary()
load_weights(filename)
save_model(modelname)
get_config()