tfl.premade_lib.compute_keypoints

Calculates keypoints for the given set of values.

tfl.premade_lib.compute_keypoints(
    values,
    num_keypoints,
    keypoints='quantiles',
    clip_min=None,
    clip_max=None,
    default_value=None,
    weights=None,
    weight_reduction='mean',
    feature_name=''
)

values Values to use for quantile calculation.
num_keypoints Number of keypoints to compute.
keypoints String 'quantiles' or 'uniform'.
clip_min Input values are lower clipped by this value.
clip_max Input values are upper clipped by this value.
default_value If provided, occurances will be removed from values.
weights Weights to be used for quantile calculation.
weight_reduction Reduction applied to weights for repeated values. Must be either 'mean' or 'sum'.
feature_name Name to use for error logs.

A list of keypoints of num_keypoints length.