From 39c20fc9535cedd0fb0b299a6173e678d0450137 Mon Sep 17 00:00:00 2001
From: Thomas Friedel <thomas.friedel@gmail.com>
Date: Sun, 26 May 2024 01:00:46 +0200
Subject: [PATCH] fix for predicting with batch_size > 1

---
 ultralytics/models/yolov10/predict.py | 29 ++++++++++++++++-----------
 1 file changed, 17 insertions(+), 12 deletions(-)

diff --git a/ultralytics/models/yolov10/predict.py b/ultralytics/models/yolov10/predict.py
index 98994f71..29f21a9c 100644
--- a/ultralytics/models/yolov10/predict.py
+++ b/ultralytics/models/yolov10/predict.py
@@ -11,28 +11,33 @@ class YOLOv10DetectionPredictor(DetectionPredictor):
 
         if isinstance(preds, (list, tuple)):
             preds = preds[0]
-        
-        if preds.shape[-1] == 6:
-            pass
-        else:
+
+        if preds.shape[-1] != 6:
             preds = preds.transpose(-1, -2)
-            bboxes, scores, labels = ops.v10postprocess(preds, self.args.max_det, preds.shape[-1]-4)
+            bboxes, scores, labels = ops.v10postprocess(
+                preds, self.args.max_det, preds.shape[-1] - 4
+            )
             bboxes = ops.xywh2xyxy(bboxes)
-            preds = torch.cat([bboxes, scores.unsqueeze(-1), labels.unsqueeze(-1)], dim=-1)
+            preds = torch.cat(
+                [bboxes, scores.unsqueeze(-1), labels.unsqueeze(-1)], dim=-1
+            )
 
         mask = preds[..., 4] > self.args.conf
 
-        b, _, c = preds.shape
-        preds = preds.view(-1, preds.shape[-1])[mask.view(-1)]
-        preds = preds.view(b, -1, c)
+        # Filter predictions using the mask and keep batch dimension
+        filtered_preds = [p[mask[idx]] for idx, p in enumerate(preds)]
 
-        if not isinstance(orig_imgs, list):  # input images are a torch.Tensor, not a list
+        if not isinstance(
+            orig_imgs, list
+        ):  # input images are a torch.Tensor, not a list
             orig_imgs = ops.convert_torch2numpy_batch(orig_imgs)
 
         results = []
-        for i, pred in enumerate(preds):
+        for i, pred in enumerate(filtered_preds):
             orig_img = orig_imgs[i]
             pred[:, :4] = ops.scale_boxes(img.shape[2:], pred[:, :4], orig_img.shape)
             img_path = self.batch[0][i]
-            results.append(Results(orig_img, path=img_path, names=self.model.names, boxes=pred))
+            results.append(
+                Results(orig_img, path=img_path, names=self.model.names, boxes=pred)
+            )
         return results