use xyxy2xysr ops

boxmot/trackers/deepocsort/deep_ocsort.py

@@ -11,6 +11,7 @@
 from boxmot.utils.iou import get_asso_func
 from boxmot.trackers.basetracker import BaseTracker
 from boxmot.utils import PerClassDecorator
+from boxmot.utils.ops import xyxy2xysr
 
 
 def k_previous_obs(observations, cur_age, k):
@@ -24,34 +25,6 @@ def k_previous_obs(observations, cur_age, k):
     return observations[max_age]
 
 
-def convert_bbox_to_z(bbox):
-    """
-    Takes a bounding box in the form [x1,y1,x2,y2] and returns z in the form
-      [x,y,s,r] where x,y is the centre of the box and s is the scale/area and r is
-      the aspect ratio
-    """
-    w = bbox[2] - bbox[0]
-    h = bbox[3] - bbox[1]
-    x = bbox[0] + w / 2.0
-    y = bbox[1] + h / 2.0
-    s = w * h  # scale is just area
-    r = w / float(h + 1e-6)
-    return np.array([x, y, s, r]).reshape((4, 1))
-
-
-def convert_bbox_to_z_new(bbox):
-    w = bbox[2] - bbox[0]
-    h = bbox[3] - bbox[1]
-    x = bbox[0] + w / 2.0
-    y = bbox[1] + h / 2.0
-    return np.array([x, y, w, h]).reshape((4, 1))
-
-
-def convert_x_to_bbox_new(x):
-    x, y, w, h = x.reshape(-1)[:4]
-    return np.array([x - w / 2, y - h / 2, x + w / 2, y + h / 2]).reshape(1, 4)
-
-
 def convert_x_to_bbox(x, score=None):
     """
     Takes a bounding box in the centre form [x,y,s,r] and returns it in the form
@@ -118,7 +91,7 @@ def __init__(self, det, delta_t=3, emb=None, alpha=0, max_obs=50):
         self.kf.P *= 10.0
         self.kf.Q[-1, -1] *= 0.01
         self.kf.Q[4:, 4:] *= 0.01
-        self.bbox_to_z_func = convert_bbox_to_z
+        self.bbox_to_z_func = xyxy2xysr
         self.x_to_bbox_func = convert_x_to_bbox
 
         self.kf.x[:4] = self.bbox_to_z_func(bbox)

boxmot/trackers/ocsort/ocsort.py

@@ -13,6 +13,7 @@
 from boxmot.utils.iou import run_asso_func
 from boxmot.trackers.basetracker import BaseTracker
 from boxmot.utils import PerClassDecorator
+from boxmot.utils.ops import xyxy2xysr
 
 
 def k_previous_obs(observations, cur_age, k):
@@ -26,21 +27,6 @@ def k_previous_obs(observations, cur_age, k):
     return observations[max_age]
 
 
-def convert_bbox_to_z(bbox):
-    """
-    Takes a bounding box in the form [x1,y1,x2,y2] and returns z in the form
-      [x,y,s,r] where x,y is the centre of the box and s is the scale/area and r is
-      the aspect ratio
-    """
-    w = bbox[2] - bbox[0]
-    h = bbox[3] - bbox[1]
-    x = bbox[0] + w / 2.0
-    y = bbox[1] + h / 2.0
-    s = w * h  # scale is just area
-    r = w / float(h + 1e-6)
-    return np.array([x, y, s, r]).reshape((4, 1))
-
-
 def convert_x_to_bbox(x, score=None):
     """
     Takes a bounding box in the centre form [x,y,s,r] and returns it in the form
@@ -109,7 +95,7 @@ def __init__(self, bbox, cls, det_ind, delta_t=3, max_obs=50):
         self.kf.Q[-1, -1] *= 0.01
         self.kf.Q[4:, 4:] *= 0.01
 
-        self.kf.x[:4] = convert_bbox_to_z(bbox)
+        self.kf.x[:4] = xyxy2xysr(bbox)
         self.time_since_update = 0
         self.id = KalmanBoxTracker.count
         KalmanBoxTracker.count += 1
@@ -165,7 +151,7 @@ def update(self, bbox, cls, det_ind):
             self.time_since_update = 0
             self.hits += 1
             self.hit_streak += 1
-            self.kf.update(convert_bbox_to_z(bbox))
+            self.kf.update(xyxy2xysr(bbox))
         else:
             self.kf.update(bbox)
 

boxmot/utils/ops.py

@@ -95,3 +95,27 @@ def tlwh2xyah(x):
     y[..., 2] = x[..., 2] / x[..., 3]
     y[..., 3] = x[..., 3]
     return y
+
+
+def xyxy2xysr(x):
+    """
+    Converts bounding box coordinates from (x1, y1, x2, y2) format to (x, y, s, r) format.
+
+    Args:
+        bbox (np.ndarray) or (torch.Tensor): The input bounding box coordinates in (x1, y1, x2, y2) format.
+    Returns:
+        z (np.ndarray) or (torch.Tensor): The bounding box coordinates in (x, y, s, r) format, where
+                                          x, y is the center of the box,
+                                          s is the scale (area), and
+                                          r is the aspect ratio.
+    """
+    x = x[0:4]
+    y = x.clone() if isinstance(x, torch.Tensor) else np.copy(x)
+    w = y[..., 2] - y[..., 0]  # width
+    h = y[..., 3] - y[..., 1]  # height
+    y[..., 0] = y[..., 0] + w / 2.0            # x center
+    y[..., 1] = y[..., 1] + h / 2.0            # y center
+    y[..., 2] = w * h                                  # scale (area)
+    y[..., 3] = w / (h + 1e-6)                         # aspect ratio
+    y = y.reshape((4, 1))
+    return y