Fix #6550 (#6551)

Changes image size convention to (h, w). I don't think this is used by
models other than SSD which assumes this convention.
Co-authored-by: Nicolas Granger <nicolas.granger@cea.fr>
Co-authored-by: Vasilis Vryniotis <datumbox@users.noreply.github.com>

torchvision/models/detection/anchor_utils.py

@@ -213,8 +213,8 @@ class DefaultBoxGenerator(nn.Module):
         for k, f_k in enumerate(grid_sizes):
             # Now add the default boxes for each width-height pair
             if self.steps is not None:
-                x_f_k = image_size[0] / self.steps[k]
-                y_f_k = image_size[1] / self.steps[k]
+                x_f_k = image_size[1] / self.steps[k]
+                y_f_k = image_size[0] / self.steps[k]
             else:
                 y_f_k, x_f_k = f_k