Entropy

class xaitk_saliency.impls.saliency_metric.entropy.Entropy(clip_range: tuple[float, float] | None = None)

Implementation of the SaliencyMetric interface to calculate entropy.

Entropy is a statistical measure of randomness that quantifies the amount of information in a saliency map. High entropy indicates complex, detailed information, while low entropy suggests uniform or less informative regions.

Attributes:

clip_range (tuple[float, float] | None): Optional (min, max) range to clip: saliency values before computing entropy.

Methods

`compute`	Compute the entropy of a saliency map.
`from_config`	Instantiate a new instance of this class given the configuration JSON-compliant dictionary encapsulating initialization arguments.
`get_config`	Generates a configuration dictionary for the Entropy metric instance.
`get_default_config`	Generate and return a default configuration dictionary for this class.
`get_impls`	Discover and return a set of classes that implement the calling class.
`is_usable`	Check whether this class is available for use.

__init__(clip_range: tuple[float, float] | None = None) → None

Initialize the Entropy metric.

Args:

clip_range (tuple[float, float] | None): Optional (min, max) value range to: clip the saliency map before computing entropy. The metric computation does a min-max normalization if clip_range is not provided. If input already contains a [-1,1] normalized saliency map, the following clip ranges can be used: - Full Saliency: clip_range = (-1, 1) - Positive Saliency: clip_range = (0, 1) - Negative Saliency: clip_range = (-1, 0)

compute(sal_map: ndarray) → float

Compute the entropy of a saliency map.

Entropy is computed using scipy.stats.entropy on the clipped/normalized saliency map. If clip_range is provided, the metric computation only performs range clipping and does not perform normalization. However, if clip_range is not provided, the metric computation performs only a min-max normalization.

Args:: sal_map (np.ndarray): The input saliency map.
Returns:: float: The entropy value of the saliency map.
Warnings:: UserWarning: If ref_sal_map is provided, it is ignored for entropy computation.

get_config() → dict[str, Any]

Generates a configuration dictionary for the Entropy metric instance.

Returns:: dict[str, Any]: Configuration data representing the sensor and scenario.