Deep heatmap-based landmark regression stage

deepLandmarkRegressionWithHeatmaps(
  model,
  activation = c("none", "relu", "trelu", "softmax", "sigmoid"),
  theta
)

Arguments

model	input deep model, presumably a unet. it should have an input dimensionality that is dimension-appropriate. the number of output channels should match the number of landmarks and should be an image.
activation	the activation function for the regression maps
theta	the theta parameter for thresholded relu

the augmented model