das.event_utils#

Utilities for handling events.

das.event_utils.detect_events(event_probability: numpy.ndarray, thres: float = 0.5, min_dist: int = 100, index: int = 0) → Tuple[numpy.ndarray, numpy.ndarray][source]#

Detect events as peaks in probabilitiy.

Parameters

  • event_probability ([np.ndarray]) – [T, nb_classes]

  • thres (float, optional) – [description]. Defaults to 0.5.

  • min_dist (int, optional) – [description]. Defaults to 100 samples.

  • index – (int, Optional): List of indices into axis 1 for which to compute the labels. Defaults to None (use all indices).

Returns

index of each detected event event_confidence: event_probability at the event_index

Return type

event_indices

das.event_utils.evaluate_eventtimes(eventtimes_true, eventtimes_pred, samplerate, tol=0.01)[source]#

[summary]

Parameters

  • eventtimes_true ([type]) – in seconds

  • eventtimes_pred ([type]) – in seconds

  • samplerate (float) – in Hz

  • tol (int, optional) – in seconds [description]. Defaults to 0.01 seconds.

Returns

[description]

Return type

[type]

das.event_utils.event_interval_filter(events: Iterable, event_dist_min: float = 0, event_dist_max: float = inf) → Iterable[source]#

[summary]

Parameters

  • events (Iterable) – Iterable (list, np.array) of event times in seconds.

  • event_dist_min (float, optional) – [description]. Defaults to 0 second.

  • event_dist_max (float, optional) – [description]. Defaults to np.inf seconds.

Returns

indices of events to keep events[good_event_indices].

Return type

good_event_indices (Iterable)

das.event_utils.find_nearest(array, values)[source]#

Find nearest occurrence of each item of values in array.

Parameters

  • array – find nearest in this list

  • values – queries

Returns

nearest val in array to each item in values idx: index of nearest val in array to each item in values dist: distance to nearest val in array for each item in values NOTE: Returns nan-arrays of the same size as values if array is empty.

Return type

val

das.event_utils.match_events(eventindices_true, eventindices_pred, tol=100)[source]#

Find events eventindices_pred that match those (within tol) in eventindices_true.

Parameters

  • eventindices_true – list of reference event indices

  • eventindices_pred – list of detected event indices

  • tol – n samples within which events are deemed identical

Returns

masked array copy of eventindices_pred, mask=True indicates entries in pred not closest within tol in true nearest_dist: dist of each eventindices_pred to the nearest true_event

Return type

nearest_event