Linker

class byotrack.api.linker.Linker

Bases: ABC

Base class for linkers in videos.

A linker solves the tracking as the optimal links of detections through time.

abstractmethod run(video: Sequence[np.ndarray] | np.ndarray | None, detections_sequence: Sequence[byotrack.DetectionsLike]) → Collection[byotrack.Track]

Run the linker on a whole sequence.

Parameters:

• video (Sequence[np.ndarray] | np.ndarray | None) – Optional video sequence of T frames. Each frame (array) is expected to have a shape ([D, ]H, W, C). Some linkers may not require any video. In that case, you may provide explicitly None.

• detections_sequence (Sequence[byotrack.DetectionsLike]) – Detections are expected for for each frame of the video (sorted in time). Note that for a given frame, the Detections can be empty (i.e. Detections.length = 0).

Returns:

Tracks for the given detections.

Return type:

Collection[byotrack.Track]

class byotrack.api.linker.OnlineLinker

Bases: Linker

Online linker. Build tracks one frame at a time.

The linking algorithm is dividing into three steps:

• reset(): Reset the algorithm.

• update(): Update the algorithm with a new frame and its detections.

• collect(): Collect the constructed tracks up to the last frame.

reset(dim=2) → None

Reset the linking algorithm.

Flush all data stored from a previous linking and prepare a new linking.

Parameters:

dim (int) – The dimension of the data. Default: 2

abstractmethod update(frame: np.ndarray | None, detections: byotrack.Detections) → None

Progress in the linking step by one frame.

Will update the internal algorithm by a single frame and its detections.

Parameters:

• frame (np.ndarray | None) – Optional frame of the video. Shape: ([D, ]H, W, C), dtype: float

• detections (byotrack.Detections) – Detections for the given frame.

abstractmethod collect() → Collection[byotrack.Track]

Collect and build all the tracks up to the last given frame.

Returns:

Tracks from the last reset to the last given frame.

Return type:

Collection[byotrack.Track]

run(video: Sequence[np.ndarray] | np.ndarray | None, detections_sequence: Sequence[byotrack.DetectionsLike]) → Collection[byotrack.Track]

Run the linker on a whole sequence.

Parameters:

• video (Sequence[np.ndarray] | np.ndarray | None) – Optional video sequence of T frames. Each frame (array) is expected to have a shape ([D, ]H, W, C). Some linkers may not require any video. In that case, you may provide explicitly None.

• detections_sequence (Sequence[byotrack.DetectionsLike]) – Detections are expected for for each frame of the video (sorted in time). Note that for a given frame, the Detections can be empty (i.e. Detections.length = 0).

Returns:

Tracks for the given detections.

Return type:

Collection[byotrack.Track]