sleap.nn.identity

Utilities for models that learn identity.

These functions implement the inference logic for classifying peaks using class maps or classification vectors.

sleap.nn.identity.classify_peaks_from_maps(class_maps: Tensor, peak_points: Tensor, peak_vals: Tensor, peak_sample_inds: Tensor, peak_channel_inds: Tensor, n_channels: int) → Tuple[Tensor, Tensor, Tensor]

Classify and group local peaks by their class map probability.

Parameters:

• class_maps – Class maps for a batch as a tf.Tensor of dtype tf.float32 and shape (n_samples, height, width, n_classes).

• peak_points – Local peak coordinates as a tf.Tensor of dtype tf.float32 and shape (n_peaks,). These should be in the same scale as the class maps.

• peak_vals – Confidence map value each peak as a tf.Tensor of dtype tf.float32 and shape (n_peaks,).

• peak_sample_inds – Sample index for each peak as a tf.Tensor of dtype tf.int32 and shape (n_peaks,).

• peak_channel_inds – Channel index for each peak as a tf.Tensor of dtype tf.int32 and shape (n_peaks,).

• n_channels – Integer number of channels (nodes) the instances should have.

Returns:

A tuple of (points, point_vals, class_probs) containing the grouped peaks.

points: Class-grouped peaks as a tf.Tensor of dtype tf.float32 and shape

(n_samples, n_classes, n_channels, 2). Missing points will be denoted by NaNs.

point_vals: The confidence map values for each point as a tf.Tensor of dtype

tf.float32 and shape (n_samples, n_classes, n_channels).

class_probs: Classification probabilities for matched points as a tf.Tensor

of dtype tf.float32 and shape (n_samples, n_classes, n_channels).

See also: group_class_peaks

sleap.nn.identity.classify_peaks_from_vectors(peak_points: Tensor, peak_vals: Tensor, peak_class_probs: Tensor, crop_sample_inds: Tensor, n_samples: int) → Tuple[Tensor, Tensor, Tensor]

Group peaks by classification probabilities.

This is used in top-down classification models.

Parameters:

• peak_points

• peak_vals

• peak_class_probs

• crop_sample_inds

• n_samples – Number of samples in the batch.

Returns:

A tuple of (points, point_vals, class_probs).

points: Class-grouped peaks as a tf.Tensor of dtype tf.float32 and shape

(n_samples, n_classes, n_channels, 2). Missing points will be denoted by NaNs.

point_vals: The confidence map values for each point as a tf.Tensor of dtype

tf.float32 and shape (n_samples, n_classes, n_channels).

class_probs: Classification probabilities for matched points as a tf.Tensor

of dtype tf.float32 and shape (n_samples, n_classes, n_channels).

sleap.nn.identity.group_class_peaks(peak_class_probs: Tensor, peak_sample_inds: Tensor, peak_channel_inds: Tensor, n_samples: int, n_channels: int) → Tuple[Tensor, Tensor]

Group local peaks using class probabilities.

This is useful for matching peaks that span multiple samples and channels into classes (e.g., instance identities) by their class probability.

Parameters:

• peak_class_probs – Class probabilities for each peak as tf.Tensor of dtype tf.float32 and shape (n_peaks, n_classes).

• peak_sample_inds – Sample index for each peak as tf.Tensor of dtype tf.int32 and shape (n_peaks,).

• peak_channel_inds – Channel index for each peak as tf.Tensor of dtype tf.int32 and shape (n_peaks,).

• n_samples – Integer number of samples in the batch.

• n_channels – Integer number of channels (nodes) the instances should have.

Returns:

A tuple of (peak_inds, class_inds).

peak_inds: Indices of class-grouped peaks within [0, n_peaks). Will be at

most n_classes long.

class_inds: Indices of the corresponding class for each peak within

[0, n_peaks). Will be at most n_classes long.

Notes

Peaks will be assigned to classes by their probability using the Hungarian algorithm. Peaks that are assigned to classes that are not the highest probability for each class are removed from the matches.

See also: classify_peaks_from_maps, classify_peaks_from_vectors