Skip to content

track_instance

sleap_nn.tracking.track_instance

TrackInstance Data structure for Tracker queue.

Classes:

Name Description
TrackInstanceLocalQueue

Data structure for instances in tracker queue for Local Queue method.
TrackInstances

Data structure for instances in tracker queue for fixed window method.
TrackedInstanceFeature

Data structure for tracked instances.

TrackInstanceLocalQueue

Data structure for instances in tracker queue for Local Queue method.

Source code in sleap_nn/tracking/track_instance.py 
21
22
23
24
25
26
27
28
29
30
31
@attrs.define
class TrackInstanceLocalQueue:
    """Data structure for instances in tracker queue for Local Queue method."""

    src_instance: sio.PredictedInstance
    src_instance_idx: int
    feature: np.array
    track_id: Optional[int] = None
    tracking_score: Optional[float] = None
    frame_idx: Optional[float] = None
    image: Optional[np.array] = None

TrackInstances

Data structure for instances in tracker queue for fixed window method.

Source code in sleap_nn/tracking/track_instance.py 
 9
10
11
12
13
14
15
16
17
18
@attrs.define
class TrackInstances:
    """Data structure for instances in tracker queue for fixed window method."""

    src_instances: List[sio.PredictedInstance]
    features: List[np.array]
    track_ids: Optional[List[int]] = None
    tracking_scores: Optional[List[float]] = None
    frame_idx: Optional[float] = None
    image: Optional[np.array] = None

TrackedInstanceFeature

Data structure for tracked instances.

This data structure is used for updating the previous tracked instances and get the features of the tracked instances. shifted_keypoints is used only for the FlowShiftTracker to store the optical flow shifted instances.

Source code in sleap_nn/tracking/track_instance.py 
34
35
36
37
38
39
40
41
42
43
44
45
46
47
@attrs.define
class TrackedInstanceFeature:
    """Data structure for tracked instances.

    This data structure is used for updating the previous tracked instances and get the
    features of the tracked instances. `shifted_keypoints` is used only for the `FlowShiftTracker`
    to store the optical flow shifted instances.
    """

    feature: np.ndarray
    src_predicted_instance: sio.PredictedInstance
    frame_idx: int
    tracking_score: float
    shifted_keypoints: np.ndarray = None