sentipy.lib package¶
Submodules¶
sentipy.lib.activations module¶
sentipy.lib.neuralnet module¶
-
class
sentipy.lib.neuralnet.
Network
(hidden_layers: Tuple[List[sentipy.lib.neuralnet.Neuron]], output_neuron: sentipy.lib.neuralnet.Neuron)[source]¶ Bases:
object
A neural network (fully-connected MLP) to implement the sentinel-2 toolbox algorithms
-
class
sentipy.lib.neuralnet.
Neuron
(weights: <sphinx.ext.autodoc.importer._MockObject object at 0x7fa368ee7110>, bias: float, activation: str)[source]¶ Bases:
object
Neuron in a Neural Network
Takes an arbitrary number of weights (depending on the number of incoming connections) activation must be one of ‘tansig’ or ‘linear’
-
calculate_potential
(input_arr: <sphinx.ext.autodoc.importer._MockObject object at 0x7fa368ee78d0>)[source]¶ Calculates the activation potential of the neuron given an input array
The length of the input array must match the number of weights (ie. the number of incoming connections)
-
forward
(input_arr: <sphinx.ext.autodoc.importer._MockObject object at 0x7fa368ee7310>) → <sphinx.ext.autodoc.importer._MockObject object at 0x7fa368ee7890>[source]¶ Completes a forward pass of the neuron, given an input array, by applying the activation function to the neuron’s activation potential.
-
sentipy.lib.preprocessing module¶
-
class
sentipy.lib.preprocessing.
Normaliser
(x_min: Union[float, <sphinx.ext.autodoc.importer._MockObject object at 0x7fa368ee7f10>], x_max: Union[float, <sphinx.ext.autodoc.importer._MockObject object at 0x7fa368eea090>])[source]¶ Bases:
object