dgenerate-ultralytics-headless 8.3.134__py3-none-any.whl

ultralytics/trackers/utils/kalman_filter.py

@@ -0,0 +1,493 @@
+# Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license
+
+import numpy as np
+import scipy.linalg
+
+
+class KalmanFilterXYAH:
+    """
+    A KalmanFilterXYAH class for tracking bounding boxes in image space using a Kalman filter.
+
+    Implements a simple Kalman filter for tracking bounding boxes in image space. The 8-dimensional state space
+    (x, y, a, h, vx, vy, va, vh) contains the bounding box center position (x, y), aspect ratio a, height h, and their
+    respective velocities. Object motion follows a constant velocity model, and bounding box location (x, y, a, h) is
+    taken as a direct observation of the state space (linear observation model).
+
+    Attributes:
+        _motion_mat (np.ndarray): The motion matrix for the Kalman filter.
+        _update_mat (np.ndarray): The update matrix for the Kalman filter.
+        _std_weight_position (float): Standard deviation weight for position.
+        _std_weight_velocity (float): Standard deviation weight for velocity.
+
+    Methods:
+        initiate: Creates a track from an unassociated measurement.
+        predict: Runs the Kalman filter prediction step.
+        project: Projects the state distribution to measurement space.
+        multi_predict: Runs the Kalman filter prediction step (vectorized version).
+        update: Runs the Kalman filter correction step.
+        gating_distance: Computes the gating distance between state distribution and measurements.
+
+    Examples:
+        Initialize the Kalman filter and create a track from a measurement
+        >>> kf = KalmanFilterXYAH()
+        >>> measurement = np.array([100, 200, 1.5, 50])
+        >>> mean, covariance = kf.initiate(measurement)
+        >>> print(mean)
+        >>> print(covariance)
+    """
+
+    def __init__(self):
+        """
+        Initialize Kalman filter model matrices with motion and observation uncertainty weights.
+
+        The Kalman filter is initialized with an 8-dimensional state space (x, y, a, h, vx, vy, va, vh), where (x, y)
+        represents the bounding box center position, 'a' is the aspect ratio, 'h' is the height, and their respective
+        velocities are (vx, vy, va, vh). The filter uses a constant velocity model for object motion and a linear
+        observation model for bounding box location.
+
+        Examples:
+            Initialize a Kalman filter for tracking:
+            >>> kf = KalmanFilterXYAH()
+        """
+        ndim, dt = 4, 1.0
+
+        # Create Kalman filter model matrices
+        self._motion_mat = np.eye(2 * ndim, 2 * ndim)
+        for i in range(ndim):
+            self._motion_mat[i, ndim + i] = dt
+        self._update_mat = np.eye(ndim, 2 * ndim)
+
+        # Motion and observation uncertainty are chosen relative to the current state estimate
+        self._std_weight_position = 1.0 / 20
+        self._std_weight_velocity = 1.0 / 160
+
+    def initiate(self, measurement: np.ndarray):
+        """
+        Create a track from an unassociated measurement.
+
+        Args:
+            measurement (np.ndarray): Bounding box coordinates (x, y, a, h) with center position (x, y), aspect ratio a,
+                and height h.
+
+        Returns:
+            (np.ndarray): Mean vector (8-dimensional) of the new track. Unobserved velocities are initialized to 0 mean.
+            (np.ndarray): Covariance matrix (8x8 dimensional) of the new track.
+
+        Examples:
+            >>> kf = KalmanFilterXYAH()
+            >>> measurement = np.array([100, 50, 1.5, 200])
+            >>> mean, covariance = kf.initiate(measurement)
+        """
+        mean_pos = measurement
+        mean_vel = np.zeros_like(mean_pos)
+        mean = np.r_[mean_pos, mean_vel]
+
+        std = [
+            2 * self._std_weight_position * measurement[3],
+            2 * self._std_weight_position * measurement[3],
+            1e-2,
+            2 * self._std_weight_position * measurement[3],
+            10 * self._std_weight_velocity * measurement[3],
+            10 * self._std_weight_velocity * measurement[3],
+            1e-5,
+            10 * self._std_weight_velocity * measurement[3],
+        ]
+        covariance = np.diag(np.square(std))
+        return mean, covariance
+
+    def predict(self, mean: np.ndarray, covariance: np.ndarray):
+        """
+        Run Kalman filter prediction step.
+
+        Args:
+            mean (np.ndarray): The 8-dimensional mean vector of the object state at the previous time step.
+            covariance (np.ndarray): The 8x8-dimensional covariance matrix of the object state at the previous time step.
+
+        Returns:
+            (np.ndarray): Mean vector of the predicted state. Unobserved velocities are initialized to 0 mean.
+            (np.ndarray): Covariance matrix of the predicted state.
+
+        Examples:
+            >>> kf = KalmanFilterXYAH()
+            >>> mean = np.array([0, 0, 1, 1, 0, 0, 0, 0])
+            >>> covariance = np.eye(8)
+            >>> predicted_mean, predicted_covariance = kf.predict(mean, covariance)
+        """
+        std_pos = [
+            self._std_weight_position * mean[3],
+            self._std_weight_position * mean[3],
+            1e-2,
+            self._std_weight_position * mean[3],
+        ]
+        std_vel = [
+            self._std_weight_velocity * mean[3],
+            self._std_weight_velocity * mean[3],
+            1e-5,
+            self._std_weight_velocity * mean[3],
+        ]
+        motion_cov = np.diag(np.square(np.r_[std_pos, std_vel]))
+
+        mean = np.dot(mean, self._motion_mat.T)
+        covariance = np.linalg.multi_dot((self._motion_mat, covariance, self._motion_mat.T)) + motion_cov
+
+        return mean, covariance
+
+    def project(self, mean: np.ndarray, covariance: np.ndarray):
+        """
+        Project state distribution to measurement space.
+
+        Args:
+            mean (np.ndarray): The state's mean vector (8 dimensional array).
+            covariance (np.ndarray): The state's covariance matrix (8x8 dimensional).
+
+        Returns:
+            (np.ndarray): Projected mean of the given state estimate.
+            (np.ndarray): Projected covariance matrix of the given state estimate.
+
+        Examples:
+            >>> kf = KalmanFilterXYAH()
+            >>> mean = np.array([0, 0, 1, 1, 0, 0, 0, 0])
+            >>> covariance = np.eye(8)
+            >>> projected_mean, projected_covariance = kf.project(mean, covariance)
+        """
+        std = [
+            self._std_weight_position * mean[3],
+            self._std_weight_position * mean[3],
+            1e-1,
+            self._std_weight_position * mean[3],
+        ]
+        innovation_cov = np.diag(np.square(std))
+
+        mean = np.dot(self._update_mat, mean)
+        covariance = np.linalg.multi_dot((self._update_mat, covariance, self._update_mat.T))
+        return mean, covariance + innovation_cov
+
+    def multi_predict(self, mean: np.ndarray, covariance: np.ndarray):
+        """
+        Run Kalman filter prediction step for multiple object states (Vectorized version).
+
+        Args:
+            mean (np.ndarray): The Nx8 dimensional mean matrix of the object states at the previous time step.
+            covariance (np.ndarray): The Nx8x8 covariance matrix of the object states at the previous time step.
+
+        Returns:
+            (np.ndarray): Mean matrix of the predicted states with shape (N, 8).
+            (np.ndarray): Covariance matrix of the predicted states with shape (N, 8, 8).
+
+        Examples:
+            >>> mean = np.random.rand(10, 8)  # 10 object states
+            >>> covariance = np.random.rand(10, 8, 8)  # Covariance matrices for 10 object states
+            >>> predicted_mean, predicted_covariance = kalman_filter.multi_predict(mean, covariance)
+        """
+        std_pos = [
+            self._std_weight_position * mean[:, 3],
+            self._std_weight_position * mean[:, 3],
+            1e-2 * np.ones_like(mean[:, 3]),
+            self._std_weight_position * mean[:, 3],
+        ]
+        std_vel = [
+            self._std_weight_velocity * mean[:, 3],
+            self._std_weight_velocity * mean[:, 3],
+            1e-5 * np.ones_like(mean[:, 3]),
+            self._std_weight_velocity * mean[:, 3],
+        ]
+        sqr = np.square(np.r_[std_pos, std_vel]).T
+
+        motion_cov = [np.diag(sqr[i]) for i in range(len(mean))]
+        motion_cov = np.asarray(motion_cov)
+
+        mean = np.dot(mean, self._motion_mat.T)
+        left = np.dot(self._motion_mat, covariance).transpose((1, 0, 2))
+        covariance = np.dot(left, self._motion_mat.T) + motion_cov
+
+        return mean, covariance
+
+    def update(self, mean: np.ndarray, covariance: np.ndarray, measurement: np.ndarray):
+        """
+        Run Kalman filter correction step.
+
+        Args:
+            mean (np.ndarray): The predicted state's mean vector (8 dimensional).
+            covariance (np.ndarray): The state's covariance matrix (8x8 dimensional).
+            measurement (np.ndarray): The 4 dimensional measurement vector (x, y, a, h), where (x, y) is the center
+                position, a the aspect ratio, and h the height of the bounding box.
+
+        Returns:
+            (np.ndarray): Measurement-corrected state mean.
+            (np.ndarray): Measurement-corrected state covariance.
+
+        Examples:
+            >>> kf = KalmanFilterXYAH()
+            >>> mean = np.array([0, 0, 1, 1, 0, 0, 0, 0])
+            >>> covariance = np.eye(8)
+            >>> measurement = np.array([1, 1, 1, 1])
+            >>> new_mean, new_covariance = kf.update(mean, covariance, measurement)
+        """
+        projected_mean, projected_cov = self.project(mean, covariance)
+
+        chol_factor, lower = scipy.linalg.cho_factor(projected_cov, lower=True, check_finite=False)
+        kalman_gain = scipy.linalg.cho_solve(
+            (chol_factor, lower), np.dot(covariance, self._update_mat.T).T, check_finite=False
+        ).T
+        innovation = measurement - projected_mean
+
+        new_mean = mean + np.dot(innovation, kalman_gain.T)
+        new_covariance = covariance - np.linalg.multi_dot((kalman_gain, projected_cov, kalman_gain.T))
+        return new_mean, new_covariance
+
+    def gating_distance(
+        self,
+        mean: np.ndarray,
+        covariance: np.ndarray,
+        measurements: np.ndarray,
+        only_position: bool = False,
+        metric: str = "maha",
+    ) -> np.ndarray:
+        """
+        Compute gating distance between state distribution and measurements.
+
+        A suitable distance threshold can be obtained from `chi2inv95`. If `only_position` is False, the chi-square
+        distribution has 4 degrees of freedom, otherwise 2.
+
+        Args:
+            mean (np.ndarray): Mean vector over the state distribution (8 dimensional).
+            covariance (np.ndarray): Covariance of the state distribution (8x8 dimensional).
+            measurements (np.ndarray): An (N, 4) matrix of N measurements, each in format (x, y, a, h) where (x, y) is the
+                bounding box center position, a the aspect ratio, and h the height.
+            only_position (bool): If True, distance computation is done with respect to box center position only.
+            metric (str): The metric to use for calculating the distance. Options are 'gaussian' for the squared
+                Euclidean distance and 'maha' for the squared Mahalanobis distance.
+
+        Returns:
+            (np.ndarray): Returns an array of length N, where the i-th element contains the squared distance between
+                (mean, covariance) and `measurements[i]`.
+
+        Examples:
+            Compute gating distance using Mahalanobis metric:
+            >>> kf = KalmanFilterXYAH()
+            >>> mean = np.array([0, 0, 1, 1, 0, 0, 0, 0])
+            >>> covariance = np.eye(8)
+            >>> measurements = np.array([[1, 1, 1, 1], [2, 2, 1, 1]])
+            >>> distances = kf.gating_distance(mean, covariance, measurements, only_position=False, metric="maha")
+        """
+        mean, covariance = self.project(mean, covariance)
+        if only_position:
+            mean, covariance = mean[:2], covariance[:2, :2]
+            measurements = measurements[:, :2]
+
+        d = measurements - mean
+        if metric == "gaussian":
+            return np.sum(d * d, axis=1)
+        elif metric == "maha":
+            cholesky_factor = np.linalg.cholesky(covariance)
+            z = scipy.linalg.solve_triangular(cholesky_factor, d.T, lower=True, check_finite=False, overwrite_b=True)
+            return np.sum(z * z, axis=0)  # square maha
+        else:
+            raise ValueError("Invalid distance metric")
+
+
+class KalmanFilterXYWH(KalmanFilterXYAH):
+    """
+    A KalmanFilterXYWH class for tracking bounding boxes in image space using a Kalman filter.
+
+    Implements a Kalman filter for tracking bounding boxes with state space (x, y, w, h, vx, vy, vw, vh), where
+    (x, y) is the center position, w is the width, h is the height, and vx, vy, vw, vh are their respective velocities.
+    The object motion follows a constant velocity model, and the bounding box location (x, y, w, h) is taken as a direct
+    observation of the state space (linear observation model).
+
+    Attributes:
+        _motion_mat (np.ndarray): The motion matrix for the Kalman filter.
+        _update_mat (np.ndarray): The update matrix for the Kalman filter.
+        _std_weight_position (float): Standard deviation weight for position.
+        _std_weight_velocity (float): Standard deviation weight for velocity.
+
+    Methods:
+        initiate: Creates a track from an unassociated measurement.
+        predict: Runs the Kalman filter prediction step.
+        project: Projects the state distribution to measurement space.
+        multi_predict: Runs the Kalman filter prediction step in a vectorized manner.
+        update: Runs the Kalman filter correction step.
+
+    Examples:
+        Create a Kalman filter and initialize a track
+        >>> kf = KalmanFilterXYWH()
+        >>> measurement = np.array([100, 50, 20, 40])
+        >>> mean, covariance = kf.initiate(measurement)
+        >>> print(mean)
+        >>> print(covariance)
+    """
+
+    def initiate(self, measurement: np.ndarray):
+        """
+        Create track from unassociated measurement.
+
+        Args:
+            measurement (np.ndarray): Bounding box coordinates (x, y, w, h) with center position (x, y), width, and height.
+
+        Returns:
+            (np.ndarray): Mean vector (8 dimensional) of the new track. Unobserved velocities are initialized to 0 mean.
+            (np.ndarray): Covariance matrix (8x8 dimensional) of the new track.
+
+        Examples:
+            >>> kf = KalmanFilterXYWH()
+            >>> measurement = np.array([100, 50, 20, 40])
+            >>> mean, covariance = kf.initiate(measurement)
+            >>> print(mean)
+            [100.  50.  20.  40.   0.   0.   0.   0.]
+            >>> print(covariance)
+            [[ 4.  0.  0.  0.  0.  0.  0.  0.]
+             [ 0.  4.  0.  0.  0.  0.  0.  0.]
+             [ 0.  0.  4.  0.  0.  0.  0.  0.]
+             [ 0.  0.  0.  4.  0.  0.  0.  0.]
+             [ 0.  0.  0.  0.  0.25  0.  0.  0.]
+             [ 0.  0.  0.  0.  0.  0.25  0.  0.]
+             [ 0.  0.  0.  0.  0.  0.  0.25  0.]
+             [ 0.  0.  0.  0.  0.  0.  0.  0.25]]
+        """
+        mean_pos = measurement
+        mean_vel = np.zeros_like(mean_pos)
+        mean = np.r_[mean_pos, mean_vel]
+
+        std = [
+            2 * self._std_weight_position * measurement[2],
+            2 * self._std_weight_position * measurement[3],
+            2 * self._std_weight_position * measurement[2],
+            2 * self._std_weight_position * measurement[3],
+            10 * self._std_weight_velocity * measurement[2],
+            10 * self._std_weight_velocity * measurement[3],
+            10 * self._std_weight_velocity * measurement[2],
+            10 * self._std_weight_velocity * measurement[3],
+        ]
+        covariance = np.diag(np.square(std))
+        return mean, covariance
+
+    def predict(self, mean, covariance):
+        """
+        Run Kalman filter prediction step.
+
+        Args:
+            mean (np.ndarray): The 8-dimensional mean vector of the object state at the previous time step.
+            covariance (np.ndarray): The 8x8-dimensional covariance matrix of the object state at the previous time step.
+
+        Returns:
+            (np.ndarray): Mean vector of the predicted state. Unobserved velocities are initialized to 0 mean.
+            (np.ndarray): Covariance matrix of the predicted state.
+
+        Examples:
+            >>> kf = KalmanFilterXYWH()
+            >>> mean = np.array([0, 0, 1, 1, 0, 0, 0, 0])
+            >>> covariance = np.eye(8)
+            >>> predicted_mean, predicted_covariance = kf.predict(mean, covariance)
+        """
+        std_pos = [
+            self._std_weight_position * mean[2],
+            self._std_weight_position * mean[3],
+            self._std_weight_position * mean[2],
+            self._std_weight_position * mean[3],
+        ]
+        std_vel = [
+            self._std_weight_velocity * mean[2],
+            self._std_weight_velocity * mean[3],
+            self._std_weight_velocity * mean[2],
+            self._std_weight_velocity * mean[3],
+        ]
+        motion_cov = np.diag(np.square(np.r_[std_pos, std_vel]))
+
+        mean = np.dot(mean, self._motion_mat.T)
+        covariance = np.linalg.multi_dot((self._motion_mat, covariance, self._motion_mat.T)) + motion_cov
+
+        return mean, covariance
+
+    def project(self, mean, covariance):
+        """
+        Project state distribution to measurement space.
+
+        Args:
+            mean (np.ndarray): The state's mean vector (8 dimensional array).
+            covariance (np.ndarray): The state's covariance matrix (8x8 dimensional).
+
+        Returns:
+            (np.ndarray): Projected mean of the given state estimate.
+            (np.ndarray): Projected covariance matrix of the given state estimate.
+
+        Examples:
+            >>> kf = KalmanFilterXYWH()
+            >>> mean = np.array([0, 0, 1, 1, 0, 0, 0, 0])
+            >>> covariance = np.eye(8)
+            >>> projected_mean, projected_cov = kf.project(mean, covariance)
+        """
+        std = [
+            self._std_weight_position * mean[2],
+            self._std_weight_position * mean[3],
+            self._std_weight_position * mean[2],
+            self._std_weight_position * mean[3],
+        ]
+        innovation_cov = np.diag(np.square(std))
+
+        mean = np.dot(self._update_mat, mean)
+        covariance = np.linalg.multi_dot((self._update_mat, covariance, self._update_mat.T))
+        return mean, covariance + innovation_cov
+
+    def multi_predict(self, mean, covariance):
+        """
+        Run Kalman filter prediction step (Vectorized version).
+
+        Args:
+            mean (np.ndarray): The Nx8 dimensional mean matrix of the object states at the previous time step.
+            covariance (np.ndarray): The Nx8x8 covariance matrix of the object states at the previous time step.
+
+        Returns:
+            (np.ndarray): Mean matrix of the predicted states with shape (N, 8).
+            (np.ndarray): Covariance matrix of the predicted states with shape (N, 8, 8).
+
+        Examples:
+            >>> mean = np.random.rand(5, 8)  # 5 objects with 8-dimensional state vectors
+            >>> covariance = np.random.rand(5, 8, 8)  # 5 objects with 8x8 covariance matrices
+            >>> kf = KalmanFilterXYWH()
+            >>> predicted_mean, predicted_covariance = kf.multi_predict(mean, covariance)
+        """
+        std_pos = [
+            self._std_weight_position * mean[:, 2],
+            self._std_weight_position * mean[:, 3],
+            self._std_weight_position * mean[:, 2],
+            self._std_weight_position * mean[:, 3],
+        ]
+        std_vel = [
+            self._std_weight_velocity * mean[:, 2],
+            self._std_weight_velocity * mean[:, 3],
+            self._std_weight_velocity * mean[:, 2],
+            self._std_weight_velocity * mean[:, 3],
+        ]
+        sqr = np.square(np.r_[std_pos, std_vel]).T
+
+        motion_cov = [np.diag(sqr[i]) for i in range(len(mean))]
+        motion_cov = np.asarray(motion_cov)
+
+        mean = np.dot(mean, self._motion_mat.T)
+        left = np.dot(self._motion_mat, covariance).transpose((1, 0, 2))
+        covariance = np.dot(left, self._motion_mat.T) + motion_cov
+
+        return mean, covariance
+
+    def update(self, mean, covariance, measurement):
+        """
+        Run Kalman filter correction step.
+
+        Args:
+            mean (np.ndarray): The predicted state's mean vector (8 dimensional).
+            covariance (np.ndarray): The state's covariance matrix (8x8 dimensional).
+            measurement (np.ndarray): The 4 dimensional measurement vector (x, y, w, h), where (x, y) is the center
+                position, w the width, and h the height of the bounding box.
+
+        Returns:
+            (np.ndarray): Measurement-corrected state mean.
+            (np.ndarray): Measurement-corrected state covariance.
+
+        Examples:
+            >>> kf = KalmanFilterXYWH()
+            >>> mean = np.array([0, 0, 1, 1, 0, 0, 0, 0])
+            >>> covariance = np.eye(8)
+            >>> measurement = np.array([0.5, 0.5, 1.2, 1.2])
+            >>> new_mean, new_covariance = kf.update(mean, covariance, measurement)
+        """
+        return super().update(mean, covariance, measurement)

ultralytics/trackers/utils/matching.py

@@ -0,0 +1,157 @@
+# Ultralytics 🚀 AGPL-3.0 License - https://ultralytics.com/license
+
+import numpy as np
+import scipy
+from scipy.spatial.distance import cdist
+
+from ultralytics.utils.metrics import batch_probiou, bbox_ioa
+
+try:
+    import lap  # for linear_assignment
+
+    assert lap.__version__  # verify package is not directory
+except (ImportError, AssertionError, AttributeError):
+    from ultralytics.utils.checks import check_requirements
+
+    check_requirements("lap>=0.5.12")  # https://github.com/gatagat/lap
+    import lap
+
+
+def linear_assignment(cost_matrix: np.ndarray, thresh: float, use_lap: bool = True) -> tuple:
+    """
+    Perform linear assignment using either the scipy or lap.lapjv method.
+
+    Args:
+        cost_matrix (np.ndarray): The matrix containing cost values for assignments, with shape (N, M).
+        thresh (float): Threshold for considering an assignment valid.
+        use_lap (bool): Use lap.lapjv for the assignment. If False, scipy.optimize.linear_sum_assignment is used.
+
+    Returns:
+        matched_indices (np.ndarray): Array of matched indices of shape (K, 2), where K is the number of matches.
+        unmatched_a (np.ndarray): Array of unmatched indices from the first set, with shape (L,).
+        unmatched_b (np.ndarray): Array of unmatched indices from the second set, with shape (M,).
+
+    Examples:
+        >>> cost_matrix = np.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]])
+        >>> thresh = 5.0
+        >>> matched_indices, unmatched_a, unmatched_b = linear_assignment(cost_matrix, thresh, use_lap=True)
+    """
+    if cost_matrix.size == 0:
+        return np.empty((0, 2), dtype=int), tuple(range(cost_matrix.shape[0])), tuple(range(cost_matrix.shape[1]))
+
+    if use_lap:
+        # Use lap.lapjv
+        # https://github.com/gatagat/lap
+        _, x, y = lap.lapjv(cost_matrix, extend_cost=True, cost_limit=thresh)
+        matches = [[ix, mx] for ix, mx in enumerate(x) if mx >= 0]
+        unmatched_a = np.where(x < 0)[0]
+        unmatched_b = np.where(y < 0)[0]
+    else:
+        # Use scipy.optimize.linear_sum_assignment
+        # https://docs.scipy.org/doc/scipy/reference/generated/scipy.optimize.linear_sum_assignment.html
+        x, y = scipy.optimize.linear_sum_assignment(cost_matrix)  # row x, col y
+        matches = np.asarray([[x[i], y[i]] for i in range(len(x)) if cost_matrix[x[i], y[i]] <= thresh])
+        if len(matches) == 0:
+            unmatched_a = list(np.arange(cost_matrix.shape[0]))
+            unmatched_b = list(np.arange(cost_matrix.shape[1]))
+        else:
+            unmatched_a = list(frozenset(np.arange(cost_matrix.shape[0])) - frozenset(matches[:, 0]))
+            unmatched_b = list(frozenset(np.arange(cost_matrix.shape[1])) - frozenset(matches[:, 1]))
+
+    return matches, unmatched_a, unmatched_b
+
+
+def iou_distance(atracks: list, btracks: list) -> np.ndarray:
+    """
+    Compute cost based on Intersection over Union (IoU) between tracks.
+
+    Args:
+        atracks (List[STrack] | List[np.ndarray]): List of tracks 'a' or bounding boxes.
+        btracks (List[STrack] | List[np.ndarray]): List of tracks 'b' or bounding boxes.
+
+    Returns:
+        (np.ndarray): Cost matrix computed based on IoU with shape (len(atracks), len(btracks)).
+
+    Examples:
+        Compute IoU distance between two sets of tracks
+        >>> atracks = [np.array([0, 0, 10, 10]), np.array([20, 20, 30, 30])]
+        >>> btracks = [np.array([5, 5, 15, 15]), np.array([25, 25, 35, 35])]
+        >>> cost_matrix = iou_distance(atracks, btracks)
+    """
+    if atracks and isinstance(atracks[0], np.ndarray) or btracks and isinstance(btracks[0], np.ndarray):
+        atlbrs = atracks
+        btlbrs = btracks
+    else:
+        atlbrs = [track.xywha if track.angle is not None else track.xyxy for track in atracks]
+        btlbrs = [track.xywha if track.angle is not None else track.xyxy for track in btracks]
+
+    ious = np.zeros((len(atlbrs), len(btlbrs)), dtype=np.float32)
+    if len(atlbrs) and len(btlbrs):
+        if len(atlbrs[0]) == 5 and len(btlbrs[0]) == 5:
+            ious = batch_probiou(
+                np.ascontiguousarray(atlbrs, dtype=np.float32),
+                np.ascontiguousarray(btlbrs, dtype=np.float32),
+            ).numpy()
+        else:
+            ious = bbox_ioa(
+                np.ascontiguousarray(atlbrs, dtype=np.float32),
+                np.ascontiguousarray(btlbrs, dtype=np.float32),
+                iou=True,
+            )
+    return 1 - ious  # cost matrix
+
+
+def embedding_distance(tracks: list, detections: list, metric: str = "cosine") -> np.ndarray:
+    """
+    Compute distance between tracks and detections based on embeddings.
+
+    Args:
+        tracks (List[STrack]): List of tracks, where each track contains embedding features.
+        detections (List[BaseTrack]): List of detections, where each detection contains embedding features.
+        metric (str): Metric for distance computation. Supported metrics include 'cosine', 'euclidean', etc.
+
+    Returns:
+        (np.ndarray): Cost matrix computed based on embeddings with shape (N, M), where N is the number of tracks
+            and M is the number of detections.
+
+    Examples:
+        Compute the embedding distance between tracks and detections using cosine metric
+        >>> tracks = [STrack(...), STrack(...)]  # List of track objects with embedding features
+        >>> detections = [BaseTrack(...), BaseTrack(...)]  # List of detection objects with embedding features
+        >>> cost_matrix = embedding_distance(tracks, detections, metric="cosine")
+    """
+    cost_matrix = np.zeros((len(tracks), len(detections)), dtype=np.float32)
+    if cost_matrix.size == 0:
+        return cost_matrix
+    det_features = np.asarray([track.curr_feat for track in detections], dtype=np.float32)
+    # for i, track in enumerate(tracks):
+    # cost_matrix[i, :] = np.maximum(0.0, cdist(track.smooth_feat.reshape(1,-1), det_features, metric))
+    track_features = np.asarray([track.smooth_feat for track in tracks], dtype=np.float32)
+    cost_matrix = np.maximum(0.0, cdist(track_features, det_features, metric))  # Normalized features
+    return cost_matrix
+
+
+def fuse_score(cost_matrix: np.ndarray, detections: list) -> np.ndarray:
+    """
+    Fuse cost matrix with detection scores to produce a single similarity matrix.
+
+    Args:
+        cost_matrix (np.ndarray): The matrix containing cost values for assignments, with shape (N, M).
+        detections (List[BaseTrack]): List of detections, each containing a score attribute.
+
+    Returns:
+        (np.ndarray): Fused similarity matrix with shape (N, M).
+
+    Examples:
+        Fuse a cost matrix with detection scores
+        >>> cost_matrix = np.random.rand(5, 10)  # 5 tracks and 10 detections
+        >>> detections = [BaseTrack(score=np.random.rand()) for _ in range(10)]
+        >>> fused_matrix = fuse_score(cost_matrix, detections)
+    """
+    if cost_matrix.size == 0:
+        return cost_matrix
+    iou_sim = 1 - cost_matrix
+    det_scores = np.array([det.score for det in detections])
+    det_scores = np.expand_dims(det_scores, axis=0).repeat(cost_matrix.shape[0], axis=0)
+    fuse_sim = iou_sim * det_scores
+    return 1 - fuse_sim  # fuse_cost