@@ -113,7 +113,7 @@ def get_landmarks_from_image(self, image_or_path, detected_faces=None):
113113
114114 landmarks = []
115115 for i , d in enumerate (detected_faces ):
116- center = torch .FloatTensor (
116+ center = torch .tensor (
117117 [d [2 ] - (d [2 ] - d [0 ]) / 2.0 , d [3 ] - (d [3 ] - d [1 ]) / 2.0 ])
118118 center [1 ] = center [1 ] - (d [3 ] - d [1 ]) * 0.12
119119 scale = (d [2 ] - d [0 ] + d [3 ] - d [1 ]) / self .face_detector .reference_scale
@@ -130,9 +130,9 @@ def get_landmarks_from_image(self, image_or_path, detected_faces=None):
130130 out += flip (self .face_alignment_net (flip (inp )).detach (), is_label = True )
131131 out = out .cpu ().numpy ()
132132
133- pts , pts_img = get_preds_fromhm (out , center , scale )
134- pts , pts_img = pts .view (68 , 2 ) * 4 , pts_img .view (68 , 2 )
133+ pts , pts_img = get_preds_fromhm (out , center .numpy (), scale )
135134 pts , pts_img = torch .from_numpy (pts ), torch .from_numpy (pts_img )
135+ pts , pts_img = pts .view (68 , 2 ) * 4 , pts_img .view (68 , 2 )
136136
137137 if self .landmarks_type == LandmarksType ._3D :
138138 heatmaps = np .zeros ((68 , 256 , 256 ), dtype = np .float32 )
0 commit comments