@@ -332,7 +332,7 @@ class DFBRFeatureExtractor(torch.nn.Module):
332332
333333 """
334334
335- def __init__ (self : torch . nn . Module ) -> None :
335+ def __init__ (self : DFBRFeatureExtractor ) -> None :
336336 """Initialize :class:`DFBRFeatureExtractor`."""
337337 super ().__init__ ()
338338 output_layers_id : list [str ] = ["16" , "23" , "30" ]
@@ -434,8 +434,8 @@ class DFBRegister:
434434 def __init__ (self : DFBRegister , patch_size : tuple [int , int ] = (224 , 224 )) -> None :
435435 """Initialize :class:`DFBRegister`."""
436436 self .patch_size = patch_size
437- self .x_scale : list [ float ] = []
438- self .y_scale : list [ float ] = []
437+ self .x_scale : np . ndarray
438+ self .y_scale : np . ndarray
439439 self .feature_extractor = DFBRFeatureExtractor ()
440440
441441 # Make this function private when full pipeline is implemented.
@@ -796,7 +796,7 @@ def find_points_inside_boundary(mask: np.ndarray, points: np.ndarray) -> np.ndar
796796 return PatchExtractor .filter_coordinates (
797797 mask_reader ,
798798 bbox_coord ,
799- mask .shape [:: - 1 ] ,
799+ ( mask .shape [1 ], mask . shape [ 0 ]) ,
800800 )
801801
802802 def filtering_matching_points (
@@ -1521,21 +1521,21 @@ def get_patch_dimensions(
15211521 """
15221522 width , height = size [0 ], size [1 ]
15231523
1524- x = [
1524+ x_info = [
15251525 np .linspace (1 , width , width , endpoint = True ),
15261526 np .ones (height ) * width ,
15271527 np .linspace (1 , width , width , endpoint = True ),
15281528 np .ones (height ),
15291529 ]
1530- x = np .array (list (itertools .chain .from_iterable (x )))
1530+ x = np .array (list (itertools .chain .from_iterable (x_info )))
15311531
1532- y = [
1532+ y_info = [
15331533 np .ones (width ),
15341534 np .linspace (1 , height , height , endpoint = True ),
15351535 np .ones (width ) * height ,
15361536 np .linspace (1 , height , height , endpoint = True ),
15371537 ]
1538- y = np .array (list (itertools .chain .from_iterable (y )))
1538+ y = np .array (list (itertools .chain .from_iterable (y_info )))
15391539
15401540 points = np .array ([x , y ]).transpose ()
15411541 transform = transform * [[1 , 1 , 0 ], [1 , 1 , 0 ], [1 , 1 , 1 ]] # remove translation
0 commit comments