diff --git a/mmpose/apis/inferencers/pose3d_inferencer.py b/mmpose/apis/inferencers/pose3d_inferencer.py
index 0fe66ac72b..819273af66 100644
--- a/mmpose/apis/inferencers/pose3d_inferencer.py
+++ b/mmpose/apis/inferencers/pose3d_inferencer.py
@@ -271,8 +271,8 @@ def preprocess_single(self,
                 K,
             ),
                                                      dtype=np.float32)
-            data_info['lifting_target'] = np.zeros((K, 3), dtype=np.float32)
-            data_info['lifting_target_visible'] = np.ones((K, 1),
+            data_info['lifting_target'] = np.zeros((1, K, 3), dtype=np.float32)
+            data_info['lifting_target_visible'] = np.ones((1, K, 1),
                                                           dtype=np.float32)
             data_info['camera_param'] = dict(w=width, h=height)
 
diff --git a/mmpose/datasets/transforms/formatting.py b/mmpose/datasets/transforms/formatting.py
index 132ecd8bb6..d047cff3c3 100644
--- a/mmpose/datasets/transforms/formatting.py
+++ b/mmpose/datasets/transforms/formatting.py
@@ -207,8 +207,9 @@ def transform(self, results: dict) -> dict:
         for key, packed_key in self.label_mapping_table.items():
             if key in results:
                 # For pose-lifting, store only target-related fields
-                if 'lifting_target' in results and key in {
-                        'keypoint_labels', 'keypoint_weights'
+                if 'lifting_target' in results and packed_key in {
+                        'keypoint_labels', 'keypoint_weights',
+                        'keypoints_visible'
                 }:
                     continue
                 if isinstance(results[key], list):