diff --git a/README.md b/README.md
index 3a189d2..dc2f44e 100644
--- a/README.md
+++ b/README.md
@@ -10,9 +10,9 @@ The pipeline is used to **extract the feature vector of a given image**. It firs
 
 **Input Arguments:**
 
-- image:
+- img_path:
   - the input image to be encoded
-  - supported types: `PIL.Image`
+  - supported types: `str` (path of the image)
 
 **Pipeline Output:**
 
@@ -36,11 +36,9 @@ $ pip3 install towhee
 
 ```python
 >>> from towhee import pipeline
->>> from PIL import Image
 
->>> img = Image.open('path/to/your/image')
 >>> embedding_pipeline = pipeline('towhee/image-embedding-resnet50')
->>> embedding = embedding_pipeline(img)
+>>> embedding = embedding_pipeline('path/to/your/image')
 ```
 
 ## How it works
diff --git a/image_embedding_resnet50.yaml b/image_embedding_resnet50.yaml
index ad6ff6c..c0b53a8 100644
--- a/image_embedding_resnet50.yaml
+++ b/image_embedding_resnet50.yaml
@@ -7,7 +7,7 @@ operators:
         inputs:
             -
                 df: '_start_df'
-                name: 'img'
+                name: 'img_path'
                 col: 0
         outputs:
             -
@@ -23,7 +23,7 @@ operators:
         inputs:
             -
                 df: 'image'
-                name: 'img'
+                name: 'img_path'
                 col: 0
         outputs:
             -
@@ -65,14 +65,14 @@ dataframes:
         name: '_start_df'
         columns:
             -
-                name: 'img'
-                vtype: 'PIL.Image'
+                name: 'img_path'
+                vtype: 'str'
     -
         name: 'image'
         columns:
             -
-                name: 'img'
-                vtype: 'PIL.Image'
+                name: 'img_path'
+                vtype: 'str'
     -
         name: 'image_preproc'
         columns:
diff --git a/readme_res/pipeline.png b/readme_res/pipeline.png
index 09bad6b..5dbc0d7 100644
Binary files a/readme_res/pipeline.png and b/readme_res/pipeline.png differ