diff --git a/torchvision/models/detection/keypoint_rcnn.py b/torchvision/models/detection/keypoint_rcnn.py
index 71893b3b0d0..93e966bae4b 100644
--- a/torchvision/models/detection/keypoint_rcnn.py
+++ b/torchvision/models/detection/keypoint_rcnn.py
@@ -188,16 +188,18 @@ def __init__(
         keypoint_roi_pool=None,
         keypoint_head=None,
         keypoint_predictor=None,
-        num_keypoints=17,
+        num_keypoints=None,
     ):
 
         assert isinstance(keypoint_roi_pool, (MultiScaleRoIAlign, type(None)))
         if min_size is None:
             min_size = (640, 672, 704, 736, 768, 800)
 
-        if num_classes is not None:
+        if num_keypoints is not None:
             if keypoint_predictor is not None:
-                raise ValueError("num_classes should be None when keypoint_predictor is specified")
+                raise ValueError("num_keypoints should be None when keypoint_predictor is specified")
+        else:
+            num_keypoints = 17
 
         out_channels = backbone.out_channels