Enhance flagging functionality in app.py and flagging.py by adding a subscription option and improving metadata handling for user information.

app.py

@@ -69,18 +69,22 @@ def inference(image):
 
 
 def flag_img_input(
-    image: gr.Image, name: str = "misdetection", email: str = "[email protected]"
+    image: gr.Image,
+    name_: str = "misdetection",
+    email_: str = "[email protected]",
+    subscribe_: bool = False,
 ):
     """Wrapper for flagging"""
-    logger.info("Flagging image - name: %s, email: %s", name, email)
+    logger.info("Flagging image - name: %s, email: %s, subscribe: %s", name_, email_, subscribe_)
 
     # Decode blob data if necessary
     if is_blob_data(image):
         image = decode_blob_data(image)
 
     metadata = {
-        "name": name,
-        "email": email,
+        "name": name_,
+        "email": email_,
+        "subscribe": subscribe_,
     }
 
     hf_writer.flag([image], metadata=metadata)
@@ -153,10 +157,11 @@ with gr.Blocks(theme=theme, css=css, title="SEA.AI Vision Demo") as demo:
             flag: gr.Button(FLAG_TXT, interactive=True, visible=visible),
             notice: gr.Markdown(value=NOTICE, visible=visible),
         }
-    
+
     # add hidden textbox for name and email (hacky but well...)
     name = gr.Textbox(label="name", visible=False, value="anonymous")
     email = gr.Textbox(label="email", visible=False, value="[email protected]")
+    subscribe = gr.Checkbox(label="subscribe", value=False, visible=False)
 
     # This needs to be called prior to the first call to callback.flag()
     hf_writer.setup([img_input], "flagged")
@@ -169,7 +174,7 @@ with gr.Blocks(theme=theme, css=css, title="SEA.AI Vision Demo") as demo:
         show_api=False,
     ).then(
         flag_img_input,
-        [img_input, name, email],
+        [img_input, name, email, subscribe],
         [],
         preprocess=False,
         show_api=True,

flagging.py

@@ -275,8 +275,14 @@ class HuggingFaceDatasetSaver(FlaggingCallback):
                     )
                 else:
                     row.append("")
+        type_to_dtype = {
+            bool: "boolean",
+            str: "string",
+            int: "integer",
+            float: "float",
+        }
         for key, value in metadata.items():
-            features[key] = {"dtype": "string", "_type": "Value"}
+            features[key] = {"dtype": type_to_dtype[type(value)], "_type": "Value"}
             row.append(value)
         return features, row