Update app.py

app.py

@@ -72,8 +72,8 @@ def process_image(
     image: Optional[Image.Image],
     url: Optional[str],
     model_name: str,
+    confidence_threshold: float = CONFIDENCE_THRESHOLD,
     for_json: bool = False,
-    confidence_threshold: float = CONFIDENCE_THRESHOLD
 ) -> Union[Dict, Tuple[Optional[Image.Image], Optional[pd.DataFrame], Optional[pd.DataFrame], Optional[pd.DataFrame], str]]:
     """
     Process an image for object detection or panoptic segmentation, handling Gradio and FastAPI inputs.
@@ -81,18 +81,21 @@ def process_image(
         image: PIL Image object from file upload (optional).
         url: URL of the image to process (optional).
         model_name: Name of the model to use (must be in VALID_MODELS).
-        for_json: If True, return JSON dict for API/JSON tab; else, return tuple for Gradio Home tab.
         confidence_threshold: Minimum confidence score for detection (default: 0.5).
+        for_json: If True, return JSON dict for API/JSON tab; else, return tuple for Gradio Home tab.
     Returns:
         For JSON: Dict with base64-encoded image, detected objects, and confidence scores.
         For Gradio: Tuple of (annotated image, objects DataFrame, unique objects DataFrame, properties DataFrame, error message).
     """
+    error_msg = ""
     try:
         # Validate input: ensure exactly one of image or URL is provided
         if image is None and not url:
-            return {"error": "Please provide an image or URL"} if for_json else (None, None, None, None, "Please provide an image or URL")
+            error_msg = "Please provide an image or URL"
+            return {"error": error_msg} if for_json else (None, None, None, None, error_msg)
         if image and url:
-            return {"error": "Provide either an image or URL, not both"} if for_json else (None, None, None, None, "Provide either an image or URL, not both")
+            error_msg = "Provide either an image or URL, not both"
+            return {"error": error_msg} if for_json else (None, None, None, None, error_msg)
         if model_name not in VALID_MODELS:
             error_msg = f"Invalid model: {model_name}. Choose from: {VALID_MODELS}"
             return {"error": error_msg} if for_json else (None, None, None, None, error_msg)
@@ -232,7 +235,7 @@ def process_image(
                 pd.DataFrame({"Unique Object": unique_objects, "Confidence Score": [f"{score:.2f}" for score in unique_confidences]})
                 if unique_objects else pd.DataFrame(columns=["Unique Object", "Confidence Score"])
             )
-            properties_df = pd.DataFrame([properties]) if properties else pd.DataFrame(columns=list(properties.keys()))
+            properties_df = pd.DataFrame([properties])
             return image, objects_df, unique_objects_df, properties_df, ""
     except requests.RequestException as e:
         # Handle URL fetch errors
@@ -279,7 +282,7 @@ async def detect_objects_endpoint(
             contents = await file.read()
             image = Image.open(BytesIO(contents)).convert("RGB")
         # Process image with specified parameters
-        result = process_image(image, image_url, model_name, for_json=True, confidence_threshold=confidence_threshold)
+        result = process_image(image, image_url, model_name, confidence_threshold=confidence_threshold, for_json=True)
         if "error" in result:
             raise HTTPException(status_code=400, detail=result["error"])
         return JSONResponse(content=result)
@@ -359,8 +362,7 @@ def create_gradio_ui() -> gr.Blocks:
                     )
                     # Clear all inputs and outputs
                     clear_btn.click(
-                        fn=lambda: [None, "", VALID_MODELS[0], 0.5, None, pd.DataFrame(columns=["Object", "Confidence Score"]), pd.DataFrame(columns=["Unique Object", "Confidence Score"]), pd.DataFrame(), None],
-                        inputs=None,
+                        fn=lambda: (None, "", VALID_MODELS[0], 0.5, None, pd.DataFrame(columns=["Object", "Confidence Score"]), pd.DataFrame(columns=["Unique Object", "Confidence Score"]), pd.DataFrame(columns=list(MODEL_DESCRIPTIONS.keys())), ""),
                         outputs=[image_input, image_url_input, model_choice, confidence_slider, output_image, objects_output, unique_objects_output, properties_output, error_output],
                     )
                 # JSON tab for API-like output
@@ -393,14 +395,13 @@ def create_gradio_ui() -> gr.Blocks:
                             url_output = gr.JSON(label="API Response")
                     # Process image and return JSON when Process button is clicked
                     url_submit_btn.click(
-                        fn=lambda img, url, model, conf: process_image(img, url, model, for_json=True, confidence_threshold=conf),
+                        fn=lambda img, url, model, conf: process_image(img, url, model, confidence_threshold=conf, for_json=True),
                         inputs=[image_input_json, image_url_input_json, url_model_choice, confidence_slider_json],
                         outputs=[url_output],
                     )
                     # Clear inputs and output
                     url_clear_btn.click(
-                        fn=lambda: [None, "", VALID_MODELS[0], 0.5, {}],
-                        inputs=None,
+                        fn=lambda: (None, "", VALID_MODELS[0], 0.5, {}),
                         outputs=[image_input_json, image_url_input_json, url_model_choice, confidence_slider_json, url_output],
                     )
                 # Help tab with usage instructions