revised for RLE assert (facebookresearch#679)

maskrcnn_benchmark/structures/segmentation_mask.py

@@ -57,10 +57,10 @@ def __init__(self, masks, size):
             elif isinstance(masks[0], dict) and "counts" in masks[0]:
                 # RLE interpretation
                 assert all(
-                    [(size[0], size[1]) == tuple(inst["size"]) for inst in masks]
-                )
-                masks = mask_utils.decode(masks)
-                masks = torch.tensor(masks).permute(2, 0, 1)
+                    [(size[1], size[0]) == tuple(inst["size"]) for inst in masks]
+                )  # in RLE, height come first in "size"
+                masks = mask_utils.decode(masks)  # [h, w, n]
+                masks = torch.tensor(masks).permute(2, 0, 1)  # [n, h, w]
             else:
                 RuntimeError(
                     "Type of `masks[0]` could not be interpreted: %s" % type(masks)
@@ -70,7 +70,7 @@ def __init__(self, masks, size):
             masks = masks.masks.clone()
         else:
             RuntimeError(
-                "Type of `masks` argument could not be interpreted:%s" % tpye(masks)
+                "Type of `masks` argument could not be interpreted:%s" % type(masks)
             )
 
         if len(masks.shape) == 2:
@@ -527,7 +527,7 @@ def __next__(self):
             self.iter_idx += 1
             return next_segmentation
         raise StopIteration()
-        
+
     next = __next__  # Python 2 compatibility
 
     def __repr__(self):