Improved BGR2RGB speeds (#3880)

* Update BGR2RGB ops

* speed improvements

* cleanup

models/common.py

@@ -259,7 +259,7 @@ class AutoShape(nn.Module):
             files.append(Path(f).with_suffix('.jpg').name)
             if im.shape[0] < 5:  # image in CHW
                 im = im.transpose((1, 2, 0))  # reverse dataloader .transpose(2, 0, 1)
-            im = im[:, :, :3] if im.ndim == 3 else np.tile(im[:, :, None], 3)  # enforce 3ch input
+            im = im[..., :3] if im.ndim == 3 else np.tile(im[..., None], 3)  # enforce 3ch input
             s = im.shape[:2]  # HWC
             shape0.append(s)  # image shape
             g = (size / max(s))  # gain

utils/datasets.py

@@ -218,7 +218,7 @@ class LoadImages:  # for inference
         img = letterbox(img0, self.img_size, stride=self.stride)[0]
 
         # Convert
-        img = img[:, :, ::-1].transpose(2, 0, 1)  # BGR to RGB and HWC to CHW
+        img = img.transpose((2, 0, 1))[::-1]  # HWC to CHW, BGR to RGB
         img = np.ascontiguousarray(img)
 
         return path, img, img0, self.cap
@@ -264,7 +264,7 @@ class LoadWebcam:  # for inference
         img = letterbox(img0, self.img_size, stride=self.stride)[0]
 
         # Convert
-        img = img[:, :, ::-1].transpose(2, 0, 1)  # BGR to RGB and HWC to CHW
+        img = img.transpose((2, 0, 1))[::-1]  # HWC to CHW, BGR to RGB
         img = np.ascontiguousarray(img)
 
         return img_path, img, img0, None
@@ -345,7 +345,7 @@ class LoadStreams:  # multiple IP or RTSP cameras
         img = np.stack(img, 0)
 
         # Convert
-        img = img[:, :, :, ::-1].transpose(0, 3, 1, 2)  # BGR to RGB and BHWC to BCHW
+        img = img[..., ::-1].transpose((0, 3, 1, 2))  # BGR to RGB, BHWC to BCHW
         img = np.ascontiguousarray(img)
 
         return self.sources, img, img0, None
@@ -526,7 +526,6 @@ class LoadImagesAndLabels(Dataset):  # for training/testing
             if random.random() < hyp['mixup']:
                 img, labels = mixup(img, labels, *load_mosaic(self, random.randint(0, self.n - 1)))
 
-
         else:
             # Load image
             img, (h0, w0), (h, w) = load_image(self, index)
@@ -579,7 +578,7 @@ class LoadImagesAndLabels(Dataset):  # for training/testing
             labels_out[:, 1:] = torch.from_numpy(labels)
 
         # Convert
-        img = img[:, :, ::-1].transpose(2, 0, 1)  # BGR to RGB, to 3 x img_height x img_width
+        img = img.transpose((2, 0, 1))[::-1]  # HWC to CHW, BGR to RGB
         img = np.ascontiguousarray(img)
 
         return torch.from_numpy(img), labels_out, self.img_files[index], shapes