diff --git a/README.md b/README.md
index 5876f78..3adedf6 100644
--- a/README.md
+++ b/README.md
@@ -6,12 +6,11 @@
 
 ## Desription
 
-This operator detects faces in the images by using [RetinaFace](https://arxiv.org/abs/1905.00641) Detector. It will return the bounding box positions and the confidence scores of detected faces. This repo is a adapataion from [repo](https://github.com/biubug6/Pytorch_Retinaface).
+This operator detects faces in the images by using [RetinaFace](https://arxiv.org/abs/1905.00641) Detector. It will return the bounding box positions and the confidence scores of detected faces. This repo is an adaptaion from [biubug6/Pytorch_Retinaface](https://github.com/biubug6/Pytorch_Retinaface).
 
 ## Code Example
 
-Load an image from path './dog.jpg'
-and use the pretrained RetinaFace to generate face bounding boxes. 
+Load an image from path './turing.png' and use the pretrained RetinaFace model to generate face bounding boxes and confidence scores. 
 
  *Write the pipeline in simplified style*:
 
@@ -34,7 +33,8 @@ import towhee
 towhee.glob['path']('turing.png') \
   .image_decode.cv2['path', 'img']() \
   .face_detection.retinaface['img', ('bbox','score')]() \
-  .select('img', 'bbox', 'score').show()
+  .select('img', 'bbox', 'score') \
+  .show()
 ```
 
 <img src="https://towhee.io/face-detection/retinaface/raw/branch/main/result2.png" alt="result2" style="height:60px;"/>
@@ -43,15 +43,15 @@ towhee.glob['path']('turing.png') \
 
 Create the operator via the following factory method.
 
-***ops.face_detection.retinaface()***
+***face_detection.retinaface()***
 
 ## Interface
 
-A face detection operator takes an image as input. it generates the bounding box positions and confidence scores back to ndarray.
+A face detection operator takes an image as input. It generates the bounding box positions and confidence scores back to ndarray.
 
 **Parameters:**
 
-​	***image***: *numpy.ndarray.*
+​	***img***: *numpy.ndarray.*
 
 ​	the image to detect faces.
 
@@ -59,11 +59,11 @@ A face detection operator takes an image as input. it generates the bounding box
 
 **Returns:**
 
-*numpy.ndarray*
+*List[(int, int, int, int)]*
 
 ​	The detected face bounding boxes.
 
-*numpy.ndarray*
+*List[float]*
 
 ​	The detected face bounding boxes confident scores.
 
diff --git a/result1.png b/result1.png
index f9e17d7..e29b9f8 100644
Binary files a/result1.png and b/result1.png differ
diff --git a/result2.png b/result2.png
index 90a6212..dae2e9f 100644
Binary files a/result2.png and b/result2.png differ
diff --git a/retinaface.py b/retinaface.py
index 889b5ce..11561af 100644
--- a/retinaface.py
+++ b/retinaface.py
@@ -29,7 +29,7 @@ from towhee._types import Image
 from towhee.types.arg import arg, to_image_color
 
 
-@register(output_schema=['bbox', 'score'])
+@register(output_schema=['box', 'score'])
 class Retinaface(Operator):
     """
     Retinaface
@@ -41,8 +41,11 @@ class Retinaface(Operator):
         self.model = Model()
 
     @arg(1, to_image_color('RGB') )
-    def __call__(self, image: Image):
-        img = torch.FloatTensor(numpy.asarray(to_pil(image)))
+    def __call__(self, img: Image):
+        img = torch.FloatTensor(numpy.asarray(to_pil(img)))
         bboxes, keypoints = self.model(img)
         bboxes = bboxes.cpu().detach().numpy()
-        return bboxes[:,:4].astype(int),  bboxes[:,4]
+        bl, sl = bboxes[:,:4].astype(int),  bboxes[:,4]
+        rbboxes = [tuple(box) for box in bl]
+        return rbboxes, sl.tolist()
+