Reference for ultralytics/trackers/byte_tracker.py#

class STrack(BaseTrack):
    """Single object tracking representation that uses Kalman filtering for state estimation.

    This class is responsible for storing all the information regarding individual tracklets and performs state updates
    and predictions based on Kalman filter.

    Attributes:
        shared_kalman (KalmanFilterXYAH): Shared Kalman filter used across all STrack instances for prediction.
        _tlwh (np.ndarray): Private attribute to store top-left corner coordinates and width and height of bounding box.
        kalman_filter (KalmanFilterXYAH): Instance of Kalman filter used for this particular object track.
        mean (np.ndarray): Mean state estimate vector.
        covariance (np.ndarray): Covariance of state estimate.
        is_activated (bool): Boolean flag indicating if the track has been activated.
        score (float): Confidence score of the track.
        tracklet_len (int): Length of the tracklet.
        cls (Any): Class label for the object.
        idx (int): Index of the matched detection in the current frame's detection set.
        frame_id (int): Current frame ID.
        start_frame (int): Frame where the object was first detected.
        angle (float | None): Optional angle information for oriented bounding boxes.

    Methods:
        predict: Predict the next state of the object using Kalman filter.
        multi_predict: Predict the next states for multiple tracks.
        activate: Activate a new tracklet.
        re_activate: Reactivate a previously lost tracklet.
        update: Update the state of a matched track.
        convert_coords: Convert bounding box to x-y-aspect-height format.
        tlwh_to_xyah: Convert tlwh bounding box to xyah format.

    Examples:
        Initialize and activate a new track
        >>> track = STrack(xywh=np.array([100, 200, 50, 80, 0]), score=0.9, cls="person")
        >>> track.activate(kalman_filter=KalmanFilterXYAH(), frame_id=1)
    """

    shared_kalman = KalmanFilterXYAH()

    def __init__(self, xywh: np.ndarray, score: float, cls: Any):
        """Initialize a new STrack instance.

        Args:
            xywh (np.ndarray): Bounding box in `(x, y, w, h, idx)` or `(x, y, w, h, angle, idx)` format, where (x, y) is
                the center, (w, h) are width and height, and `idx` is the detection index.
            score (float): Confidence score of the detection.
            cls (Any): Class label for the detected object.
        """
        super().__init__()
        # xywh+idx or xywha+idx
        assert len(xywh) in {5, 6}, f"expected 5 or 6 values but got {len(xywh)}"
        self._tlwh = np.asarray(xywh2ltwh(xywh[:4]), dtype=np.float32)
        self.kalman_filter = None
        self.mean, self.covariance = None, None
        self.is_activated = False

        self.score = score
        self.tracklet_len = 0
        self.cls = cls
        self.idx = xywh[-1]
        self.angle = xywh[4] if len(xywh) == 6 else None

@property
def tlwh(self) -> np.ndarray:
    """Get the bounding box in top-left-width-height format from the current state estimate."""
    if self.mean is None:
        return self._tlwh.copy()
    ret = self.mean[:4].copy()
    ret[2] *= ret[3]
    ret[:2] -= ret[2:] / 2
    return ret

@property
def xyxy(self) -> np.ndarray:
    """Get the bounding box in (min x, min y, max x, max y) format from the current state estimate."""
    ret = self.tlwh  # already a fresh array, safe to mutate
    ret[2:] += ret[:2]
    return ret

@property
def xywh(self) -> np.ndarray:
    """Get the current position of the bounding box in (center x, center y, width, height) format."""
    ret = np.asarray(self.tlwh).copy()
    ret[:2] += ret[2:] / 2
    return ret

@property
def xywha(self) -> np.ndarray:
    """Get position in (center x, center y, width, height, angle) format, warning if angle is missing."""
    if self.angle is None:
        LOGGER.warning("`angle` attr not found, returning `xywh` instead.")
        return self.xywh
    return np.concatenate([self.xywh, self.angle[None]])

@property
def result(self) -> list[float]:
    """Get the current tracking result as `[*box, track_id, score, cls, idx]`, with box in xyxy or xywha (OBB)."""
    coords = self.xyxy if self.angle is None else self.xywha
    return [*coords.tolist(), self.track_id, self.score, self.cls, self.idx]

def __repr__(self) -> str:
    """Return a string representation of the STrack object including start frame, end frame, and track ID."""
    return f"OT_{self.track_id}_({self.start_frame}-{self.end_frame})"

def activate(self, kalman_filter: KalmanFilterXYAH, frame_id: int):
    """Activate a new tracklet using the provided Kalman filter and initialize its state and covariance."""
    self.kalman_filter = kalman_filter
    self.track_id = self.next_id()
    self.mean, self.covariance = self.kalman_filter.initiate(self.convert_coords(self._tlwh))

    self.tracklet_len = 0
    self.state = TrackState.Tracked
    if frame_id == 1:
        self.is_activated = True
    self.frame_id = frame_id
    self.start_frame = frame_id

def convert_coords(self, tlwh: np.ndarray) -> np.ndarray:
    """Convert a bounding box's top-left-width-height format to its x-y-aspect-height equivalent."""
    return self.tlwh_to_xyah(tlwh)

@staticmethod
def multi_predict(stracks: list[STrack]):
    """Perform multi-object predictive tracking using Kalman filter for the provided list of STrack instances."""
    if not stracks:
        return
    multi_mean = np.asarray([st.mean for st in stracks])
    multi_covariance = np.asarray([st.covariance for st in stracks])
    for i, st in enumerate(stracks):
        if st.state != TrackState.Tracked:
            multi_mean[i][7] = 0
    multi_mean, multi_covariance = STrack.shared_kalman.multi_predict(multi_mean, multi_covariance)
    for i, (mean, cov) in enumerate(zip(multi_mean, multi_covariance)):
        stracks[i].mean = mean
        stracks[i].covariance = cov

def predict(self):
    """Predict the next state (mean and covariance) of the object using the Kalman filter."""
    mean_state = self.mean.copy()
    if self.state != TrackState.Tracked:
        mean_state[7] = 0
    self.mean, self.covariance = self.kalman_filter.predict(mean_state, self.covariance)

def re_activate(self, new_track: STrack, frame_id: int, new_id: bool = False):
    """Reactivate a previously lost track using new detection data and update its state and attributes."""
    self.mean, self.covariance = self.kalman_filter.update(
        self.mean, self.covariance, self.convert_coords(new_track.tlwh)
    )
    self.tracklet_len = 0
    self.state = TrackState.Tracked
    self.is_activated = True
    self.frame_id = frame_id
    if new_id:
        self.track_id = self.next_id()
    self.score = new_track.score
    self.cls = new_track.cls
    self.angle = new_track.angle
    self.idx = new_track.idx

@staticmethod
def tlwh_to_xyah(tlwh: np.ndarray) -> np.ndarray:
    """Convert bounding box from tlwh format to center-x-center-y-aspect-height (xyah) format."""
    ret = np.asarray(tlwh).copy()
    ret[:2] += ret[2:] / 2
    ret[2] /= ret[3]
    return ret

def update(self, new_track: STrack, frame_id: int):
    """Update the state of a matched track.

    Args:
        new_track (STrack): The new track containing updated information.
        frame_id (int): The ID of the current frame.

    Examples:
        Update the state of a track with new detection information
        >>> track = STrack(np.array([100, 200, 50, 80, 0]), score=0.9, cls=0)
        >>> track.activate(KalmanFilterXYAH(), 1)
        >>> new_track = STrack(np.array([105, 205, 55, 85, 0]), score=0.95, cls=0)
        >>> track.update(new_track, 2)
    """
    self.frame_id = frame_id
    self.tracklet_len += 1

    new_tlwh = new_track.tlwh
    self.mean, self.covariance = self.kalman_filter.update(
        self.mean, self.covariance, self.convert_coords(new_tlwh)
    )
    self.state = TrackState.Tracked
    self.is_activated = True

    self.score = new_track.score
    self.cls = new_track.cls
    self.angle = new_track.angle
    self.idx = new_track.idx

class BYTETracker:
    """BYTETracker: A tracking algorithm built on top of YOLO for object detection and tracking.

    This class encapsulates the functionality for initializing, updating, and managing the tracks for detected objects
    in a video sequence. It maintains the state of tracked, lost, and removed tracks over frames, utilizes Kalman
    filtering for predicting the new object locations, and performs data association.

    Attributes:
        tracked_stracks (list[STrack]): List of successfully activated tracks.
        lost_stracks (list[STrack]): List of lost tracks.
        removed_stracks (list[STrack]): List of removed tracks.
        removed_stracks_frame (list[STrack]): Tracks removed in the most recent frame.
        frame_id (int): The current frame ID.
        args (Namespace): Tracker configuration parsed from the tracker YAML (e.g. bytetrack.yaml).
        max_frames_lost (int): The maximum frames for a track to be considered as 'lost'.
        kalman_filter (KalmanFilterXYAH): Kalman Filter object.

    Methods:
        update: Update object tracker with new detections.
        get_kalmanfilter: Return a Kalman filter object for tracking bounding boxes.
        init_track: Initialize object tracking with detections.
        get_dists: Calculate the distance between tracks and detections.
        multi_predict: Predict the location of tracks.
        reset_id: Restart this tracker's track IDs at 1.
        reset: Reset the tracker by clearing all tracks.

    Examples:
        Initialize BYTETracker and update with detection results
        >>> from ultralytics import YOLO
        >>> from ultralytics.utils import YAML, IterableSimpleNamespace
        >>> from ultralytics.utils.checks import check_yaml
        >>> args = IterableSimpleNamespace(**YAML.load(check_yaml("bytetrack.yaml")))
        >>> tracker = BYTETracker(args)
        >>> result = YOLO("yolo26n.pt")("https://ultralytics.com/images/bus.jpg")[0]
        >>> tracked_objects = tracker.update(result.boxes.cpu().numpy(), result.orig_img)
    """

    track_class = STrack

    def __init__(self, args):
        """Initialize a BYTETracker instance for object tracking.

        Args:
            args (Namespace): Tracker configuration containing `track_high_thresh`, `track_low_thresh`,
                `new_track_thresh`, `track_buffer`, `match_thresh`, and `fuse_score`.
        """
        self.tracked_stracks: list[STrack] = []
        self.lost_stracks: list[STrack] = []
        self.removed_stracks: list[STrack] = []

        self.frame_id = 0
        self.args = args
        self.max_frames_lost = args.track_buffer
        self.kalman_filter = self.get_kalmanfilter()
        self.reset_id()

def _apply_match(self, track: STrack, det: STrack, activated: list[STrack], refind: list[STrack]) -> None:
    """Update or re-activate a single track with its matched detection."""
    if track.state == TrackState.Tracked:
        track.update(det, self.frame_id)
        activated.append(track)
    else:
        track.re_activate(det, self.frame_id, new_id=False)
        refind.append(track)

def _apply_matches(
    self,
    matches: list[list[int]] | np.ndarray,
    pool: list[STrack],
    detections: list[STrack],
    activated: list[STrack],
    refind: list[STrack],
) -> None:
    """Apply a list of matched (track, detection) pairs from an association stage."""
    for itracked, idet in matches:
        self._apply_match(pool[itracked], detections[idet], activated, refind)

def _first_association(
    self, strack_pool: list[STrack], detections: list[STrack], activated: list[STrack], refind: list[STrack]
) -> tuple[list[int], list[int]]:
    """First-stage association between track pool and high-score detections.

    Returns:
        (tuple[list[int], list[int]]): Unmatched track indices and unmatched detection indices.
    """
    dists = self.get_dists(strack_pool, detections)
    matches, u_track, u_detection = matching.linear_assignment(dists, thresh=self.args.match_thresh)
    self._apply_matches(matches, strack_pool, detections, activated, refind)
    return u_track, u_detection

def _format_output(self) -> np.ndarray:
    """Format the current tracked objects into the output array."""
    return np.asarray([x.result for x in self.tracked_stracks if x.is_activated], dtype=np.float32)

def _init_new_tracks(
    self,
    u_detection: list[int],
    detections: list[STrack],
    activated: list[STrack],
    refind: list[STrack] | None = None,
) -> None:
    """Activate new tracks from detections that survived all association stages."""
    for inew in u_detection:
        track = detections[inew]
        if track.score < self.args.new_track_thresh:
            continue
        track.activate(self.kalman_filter, self.frame_id)
        activated.append(track)

def _input_for(self, img: np.ndarray | None, feats: np.ndarray | None, mask: np.ndarray) -> Any:
    """Return the per-detection auxiliary input for ``init_track``.

    When ``feats`` is provided it is sliced by the detection mask. Trackers with a native
    (``model="auto"``) ReID encoder get None when feats are missing (e.g. user-supplied
    detections), so ``init_track`` falls back to the no-encoding path instead of feeding the
    BGR frame into the auto encoder. External ReID models always take the frame.

    Args:
        img (np.ndarray | None): Current BGR frame.
        feats (np.ndarray | None): Optional per-detection features.
        mask (np.ndarray): Boolean mask used to slice ``feats``.

    Returns:
        (Any): The auxiliary payload (features, image or None) to hand to ``init_track``.
    """
    if feats is not None and len(feats):
        return feats[mask]
    if getattr(self, "encoder", None) is not None and getattr(self.args, "model", "auto") == "auto":
        return None
    return img

def _post_first_association(
    self,
    strack_pool: list[STrack],
    detections: list[STrack],
    u_track: list[int],
    u_detection: list[int],
    activated: list[STrack],
    refind: list[STrack],
) -> tuple[list[int], list[int]]:
    """Hook executed after the first association stage and before the second.

    Returns:
        (tuple[list[int], list[int]]): Potentially modified unmatched track and detection indices.
    """
    return u_track, u_detection

def _pre_first_associate(
    self, strack_pool: list[STrack], unconfirmed: list[STrack], img: np.ndarray | None, results_high: Any
) -> None:
    """Hook called after Kalman predict, before first-stage assignment. Default: GMC if available."""
    if hasattr(self, "gmc") and self.gmc.method is not None and img is not None:
        try:
            warp = self.gmc.apply(img, results_high.xyxy)
        except Exception as e:
            LOGGER.warning(f"GMC failed, falling back to identity: {e}")
            warp = np.eye(2, 3)
        multi_gmc(strack_pool, warp)
        multi_gmc(unconfirmed, warp)

def _remove_stale_lost(self, removed: list[STrack]) -> None:
    """Remove lost tracks that have exceeded the maximum allowed frames."""
    for track in self.lost_stracks:
        if self.frame_id - track.end_frame > self.max_frames_lost:
            track.mark_removed()
            removed.append(track)

def _second_association(
    self,
    strack_pool: list[STrack],
    u_track: list[int],
    detections_second: list[STrack],
    activated: list[STrack],
    refind: list[STrack],
    lost: list[STrack],
) -> None:
    """Second-stage association between remaining tracked tracks and low-score detections."""
    r_tracked_stracks = [strack_pool[i] for i in u_track if strack_pool[i].state == TrackState.Tracked]
    if r_tracked_stracks and detections_second:
        # IoU-only by design (ByteTrack paper sec. 3.2): fusing low scores pushes costs above the 0.5 threshold
        dists = matching.iou_distance(r_tracked_stracks, detections_second)
        matches, u_track, _ = matching.linear_assignment(dists, thresh=0.5)
        self._apply_matches(matches, r_tracked_stracks, detections_second, activated, refind)
    else:
        u_track = list(range(len(r_tracked_stracks)))

    for it in u_track:
        track = r_tracked_stracks[it]
        track.mark_lost()
        lost.append(track)

def _split_detections(self, results: Any) -> tuple[Any, Any, np.ndarray, np.ndarray]:
    """Split detections into high-confidence and low-confidence subsets, dropping degenerate boxes.

    Args:
        results (Any): Results-like object with ``conf`` and ``xywh``/``xywhr`` attributes supporting boolean
            indexing.

    Returns:
        (tuple[Any, Any, np.ndarray, np.ndarray]): High-confidence results, low-confidence results, high mask, and
            low mask.
    """
    scores = results.conf
    wh = (results.xywhr if hasattr(results, "xywhr") else results.xywh)[:, 2:4]
    valid = (wh[:, 0] > 0) & (wh[:, 1] > 0)  # tlwh_to_xyah divides by height, so h=0 would give an inf Kalman mean
    remain_inds = valid & (scores >= self.args.track_high_thresh)
    inds_low = valid & (scores > self.args.track_low_thresh) & (scores < self.args.track_high_thresh)
    return results[remain_inds], results[inds_low], remain_inds, inds_low

def _split_tracked(self) -> tuple[list[STrack], list[STrack]]:
    """Separate ``self.tracked_stracks`` into confirmed and unconfirmed lists.

    Returns:
        (tuple[list[STrack], list[STrack]]): ``(unconfirmed, tracked)`` where ``unconfirmed`` holds tracks whose
            ``is_activated`` flag is False.
    """
    unconfirmed, tracked = [], []
    for track in self.tracked_stracks:
        (unconfirmed if not track.is_activated else tracked).append(track)
    return unconfirmed, tracked

def _unconfirmed_association(
    self,
    unconfirmed: list[STrack],
    u_detection: list[int],
    detections: list[STrack],
    activated: list[STrack],
    removed: list[STrack],
) -> tuple[list[int], list[STrack]]:
    """Associate unconfirmed tracks with leftover high-score detections.

    Returns:
        (tuple[list[int], list[STrack]]): Unmatched detection indices after association, and the filtered detection
            list those indices refer to.
    """
    detections = [detections[i] for i in u_detection]
    if not unconfirmed:
        return list(range(len(detections))), detections
    dists = self.get_dists(unconfirmed, detections)
    matches, u_unconfirmed, u_detection = matching.linear_assignment(dists, thresh=0.7)
    for itracked, idet in matches:
        unconfirmed[itracked].update(detections[idet], self.frame_id)
        activated.append(unconfirmed[itracked])
    for it in u_unconfirmed:
        track = unconfirmed[it]
        track.mark_removed()
        removed.append(track)
    return u_detection, detections

def get_dists(self, tracks: list[STrack], detections: list[STrack]) -> np.ndarray:
    """Calculate the distance between tracks and detections using IoU and optionally fuse scores."""
    dists = matching.iou_distance(tracks, detections)
    if self.args.fuse_score:
        dists = matching.fuse_score(dists, detections)
    return dists

def get_kalmanfilter(self) -> KalmanFilterXYAH:
    """Return a Kalman filter object for tracking bounding boxes using KalmanFilterXYAH."""
    return KalmanFilterXYAH()

def init_track(self, results, img: np.ndarray | None = None) -> list[STrack]:
    """Initialize object tracking with given detections, scores, and class labels as STrack instances."""
    if len(results) == 0:
        return []
    bboxes = parse_bboxes(results)
    return [self.track_class(xywh, s, c) for (xywh, s, c) in zip(bboxes, results.conf, results.cls)]

def multi_predict(self, tracks: list[STrack]):
    """Predict the next states for multiple tracks using Kalman filter."""
    STrack.multi_predict(tracks)

def reset(self):
    """Reset the tracker by clearing all tracked, lost, and removed tracks and reinitializing the Kalman filter."""
    self.tracked_stracks = []
    self.lost_stracks = []
    self.removed_stracks = []
    self.frame_id = 0
    self.kalman_filter = self.get_kalmanfilter()
    self.reset_id()

def reset_id(self):
    """Restart this tracker's track IDs at 1."""
    self._ids = count(1)

def update(self, results, img: np.ndarray | None = None, feats: np.ndarray | None = None, **kwargs) -> np.ndarray:
    """Update the tracker with new detections and return the current list of tracked objects.

    Args:
        results (Any): NumPy-backed detections (e.g. `Boxes` or `OBB` after `.cpu().numpy()`) exposing `conf`,
            `cls`, and `xywh` (or `xywhr`), and supporting boolean indexing.
        img (np.ndarray | None): Current BGR frame, used for GMC and external ReID models.
        feats (np.ndarray | None): Optional per-detection features for native (`model="auto"`) ReID.
        **kwargs (Any): Additional tracker-specific inputs, ignored by BYTETracker.

    Returns:
        (np.ndarray): Array of shape (N, 8) with `[x1, y1, x2, y2, track_id, score, cls, idx]` rows, or (N, 9) with
            `[x, y, w, h, angle, track_id, score, cls, idx]` rows for OBB, where `idx` is the detection index.
    """
    self.frame_id += 1
    activated_stracks = []
    refind_stracks = []
    lost_stracks = []
    removed_stracks = []

    results_high, results_low, mask_high, mask_low = self._split_detections(results)
    detections = self.init_track(results_high, self._input_for(img, feats, mask_high))
    detections_second = self.init_track(results_low, self._input_for(img, feats, mask_low))
    for tracks, mask in ((detections, mask_high), (detections_second, mask_low)):
        for track, i in zip(tracks, np.flatnonzero(mask)):
            track.idx = i  # idx must be in full detection-set space; parse_bboxes only sees the subset
            track.next_id = self._ids.__next__  # IDs are per tracker, so other trackers cannot reissue them

    unconfirmed, tracked_stracks = self._split_tracked()
    strack_pool = joint_stracks(tracked_stracks, self.lost_stracks)
    self.multi_predict(strack_pool)
    self._pre_first_associate(strack_pool, unconfirmed, img, results_high)

    u_track, u_detection = self._first_association(strack_pool, detections, activated_stracks, refind_stracks)
    u_track, u_detection = self._post_first_association(
        strack_pool, detections, u_track, u_detection, activated_stracks, refind_stracks
    )
    self._second_association(
        strack_pool, u_track, detections_second, activated_stracks, refind_stracks, lost_stracks
    )
    u_detection, detections = self._unconfirmed_association(
        unconfirmed, u_detection, detections, activated_stracks, removed_stracks
    )
    self._init_new_tracks(u_detection, detections, activated_stracks, refind_stracks)
    self._remove_stale_lost(removed_stracks)

    merge_track_pools(self, activated_stracks, refind_stracks, lost_stracks, removed_stracks)
    return self._format_output()