Faster IoU prediction matching by removing `torch.cat` (#4708)

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

ultralytics/engine/validator.py

@@ -24,6 +24,7 @@ from pathlib import Path
 
 import numpy as np
 import torch
+from scipy.optimize import linear_sum_assignment
 
 from ultralytics.cfg import get_cfg, get_save_dir
 from ultralytics.data.utils import check_cls_dataset, check_det_dataset
@@ -204,7 +205,7 @@ class BaseValidator:
                 LOGGER.info(f"Results saved to {colorstr('bold', self.save_dir)}")
             return stats
 
-    def match_predictions(self, pred_classes, true_classes, iou):
+    def match_predictions(self, pred_classes, true_classes, iou, use_scipy=False):
         """
         Matches predictions to ground truth objects (pred_classes, true_classes) using IoU.
 
@@ -212,23 +213,35 @@ class BaseValidator:
             pred_classes (torch.Tensor): Predicted class indices of shape(N,).
             true_classes (torch.Tensor): Target class indices of shape(M,).
             iou (torch.Tensor): An NxM tensor containing the pairwise IoU values for predictions and ground of truth
+            use_scipy (bool): Whether to use scipy for matching (more precise).
 
         Returns:
             (torch.Tensor): Correct tensor of shape(N,10) for 10 IoU thresholds.
         """
+        # Dx10 matrix, where D - detections, 10 - IoU thresholds
         correct = np.zeros((pred_classes.shape[0], self.iouv.shape[0])).astype(bool)
+        # LxD matrix where L - labels (rows), D - detections (columns)
         correct_class = true_classes[:, None] == pred_classes
-        for i, iouv in enumerate(self.iouv):
-            x = torch.nonzero(iou.ge(iouv) & correct_class)  # IoU > threshold and classes match
-            if x.shape[0]:
-                # Concatenate [label, detect, iou]
-                matches = torch.cat((x, iou[x[:, 0], x[:, 1]].unsqueeze(1)), 1).cpu().numpy()
-                if x.shape[0] > 1:
-                    matches = matches[matches[:, 2].argsort()[::-1]]
-                    matches = matches[np.unique(matches[:, 1], return_index=True)[1]]
-                    # matches = matches[matches[:, 2].argsort()[::-1]]
-                    matches = matches[np.unique(matches[:, 0], return_index=True)[1]]
-                correct[matches[:, 1].astype(int), i] = True
+        iou = iou * correct_class  # zero out the wrong classes
+        iou = iou.cpu().numpy()
+        for i, threshold in enumerate(self.iouv.cpu().tolist()):
+            if use_scipy:
+                cost_matrix = iou * (iou >= threshold)
+                if cost_matrix.any():
+                    labels_idx, detections_idx = linear_sum_assignment(cost_matrix, maximize=True)
+                    valid = cost_matrix[labels_idx, detections_idx] > 0
+                    if valid.any():
+                        correct[detections_idx[valid], i] = True
+            else:
+                matches = np.nonzero(iou >= threshold)  # IoU > threshold and classes match
+                matches = np.array(matches).T
+                if matches.shape[0]:
+                    if matches.shape[0] > 1:
+                        matches = matches[iou[matches[:, 0], matches[:, 1]].argsort()[::-1]]
+                        matches = matches[np.unique(matches[:, 1], return_index=True)[1]]
+                        # matches = matches[matches[:, 2].argsort()[::-1]]
+                        matches = matches[np.unique(matches[:, 0], return_index=True)[1]]
+                    correct[matches[:, 1].astype(int), i] = True
         return torch.tensor(correct, dtype=torch.bool, device=pred_classes.device)
 
     def add_callback(self, event: str, callback):