src.plot.kpconv_layer_plot

Classes

KPConvLayerPlot(**kwargs)

class src.plot.kpconv_layer_plot.KPConvLayerPlot(**kwargs)
Author:

Alberto M. Esmoris Pena

Class to plot the insights of a KPConv layer.

See KPConvLayerPlot and KPConvLayer.

Variables:

  • Q (np.ndarray) – The matrix representing the kernel’s structure.

  • W (np.ndarray) – The matrices representing the kernel’s weights.

  • Wpast (np.ndarray) – The matrices representing the kernel’s weights at a previous state.

  • sigma (float) – The influence distance of the kernel.

  • name (str) – The name of the layer containing the kernel.

__init__(**kwargs)

Initialize an instance of KPConvLayerPlot.

Parameters:

kwargs (dict) – The key-word arguments defining the plot’s attributes.

plot(**kwargs)

Plot the structure space and the matrices of weights representing the kernel of the KPConv layer.

See plot.Plot.plot().

plot_kernel_structure(**kwargs)

Plot the kernel’s structure.

Parameters:

kwargs – The key-word arguments.

Returns:

Nothing, but the plot is written to a file.

plot_kernel_weights(W, plot_name, plot_title, **kwargs)

Plot the kernel’s weights.

Parameters:

  • W – The tensor whose slices are the weights of the kernel.

  • plot_name – The name of the plot.

  • plot_title – The title of the plot.

  • kwargs – The key-word arguments.

Returns:

Nothing, but the plot is written to a file.

static plot_kernel_weights_as_matrix(fig, ax, Wk)

Assist the KPConvLayerPlot.plot_kernel_weights() function by providing the logic to plot a matrix of weights as a matrix.

static plot_kernel_weights_as_historgram(fig, ax, Wk)

Assist the KPConvLayerPlot.plot_kernel_weights() function by providing the logic to plot a matrix of weights as a histogram.