Update app.py

app.py

@@ -4,13 +4,12 @@ import numpy as np
 import pickle
 from functools import lru_cache
 
-# ---- Import your landmarks util ----
+
 try:
     from util import get_face_landmarks
 except Exception as e:
     raise ImportError(
         "Could not import 'get_face_landmarks' from util.py. "
-        "Make sure util.py exists and defines get_face_landmarks(img, draw: bool, static_image_mode: bool)."
     ) from e
 
 
@@ -38,75 +37,43 @@ def predict_emotion(image, draw_toggle):
     image: PIL.Image (from gr.Image with type='pil')
     draw_toggle: 'OFF' or 'ON'
     """
-    # Input validation
     if image is None:
         return {"Status": 1.0}, None, "Please upload an image."
 
     draw = (draw_toggle == "ON")
 
     # Convert PIL -> OpenCV BGR
-    try:
-        img_rgb = np.array(image)  # PIL -> RGB ndarray
-        if img_rgb.ndim == 2:  # grayscale to 3-ch
-            img_rgb = cv2.cvtColor(img_rgb, cv2.COLOR_GRAY2RGB)
-        img_bgr = cv2.cvtColor(img_rgb, cv2.COLOR_RGB2BGR)
-    except Exception:
-        return {"Status": 1.0}, None, "⚠ Could not read the image. Please try a different one."
-
-    # Extract landmarks (your util may also draw on img internally when draw=True)
-    try:
-        landmarks = get_face_landmarks(img_bgr, draw=draw, static_image_mode=True)
-    except Exception as e:
-        return {"Status": 1.0}, img_rgb, f"⚠ Landmark extraction failed: {e}"
-
-    # Handle no-face case (do NOT annotate; return original image)
+    img_rgb = np.array(image)
+    if img_rgb.ndim == 2:
+        img_rgb = cv2.cvtColor(img_rgb, cv2.COLOR_GRAY2RGB)
+    img_bgr = cv2.cvtColor(img_rgb, cv2.COLOR_RGB2BGR)
+
+    # Extract landmarks
+    landmarks = get_face_landmarks(img_bgr, draw=draw, static_image_mode=True)
+
+    # Handle no-face case
     if landmarks is None or (hasattr(landmarks, "__len__") and len(landmarks) == 0):
-        return {"No face detected": 1.0}, img_rgb, "⚠ No face detected in the image."
+        return {"No face detected": 1.0}, img_rgb, "No face detected in the image."
 
     # Load model
-    try:
-        model = load_model()
-    except FileNotFoundError:
-        return {"Status": 1.0}, img_rgb, "⚠ model.pkl not found in repo root."
-    except Exception as e:
-        return {"Status": 1.0}, img_rgb, f"⚠ Failed to load model: {e}"
+    model = load_model()
 
     # Predict
-    try:
-        output = model.predict([landmarks])
-        pred_idx = int(output[0])
-        pred_label = EMOTIONS[pred_idx] if 0 <= pred_idx < len(EMOTIONS) else str(pred_idx)
-
-        # Confidence/probabilities if available
-        if hasattr(model, "predict_proba"):
-            probs = model.predict_proba([landmarks])[0]
-            confidence = {EMOTIONS[i]: float(probs[i]) for i in range(len(EMOTIONS))}
-        else:
-            confidence = {pred_label: 1.0}
-
-        # Always draw predicted text on the copy we return (but ONLY if a face exists)
-        img_annot = img_bgr.copy()
-        try:
-            cv2.putText(
-                img_annot,
-                pred_label,
-                (10, img_annot.shape[0] - 10),
-                cv2.FONT_HERSHEY_SIMPLEX,
-                1.0,
-                (0, 255, 0),
-                2,
-                cv2.LINE_AA,
-            )
-        except Exception:
-            # If drawing fails, just return the original image
-            img_annot = img_bgr
+    output = model.predict([landmarks])
+    pred_idx = int(output[0])
+    pred_label = EMOTIONS[pred_idx] if 0 <= pred_idx < len(EMOTIONS) else str(pred_idx)
+
+    if hasattr(model, "predict_proba"):
+        probs = model.predict_proba([landmarks])[0]
+        confidence = {EMOTIONS[i]: float(probs[i]) for i in range(len(EMOTIONS))}
+    else:
+        confidence = {pred_label: 1.0}
 
-        img_out = cv2.cvtColor(img_annot, cv2.COLOR_BGR2RGB)
-        status = f"✅ Detected emotion: {pred_label}"
-        return confidence, img_out, status
+    # If draw_toggle is ON, landmarks drawn on img_bgr by util
+    img_out = cv2.cvtColor(img_bgr, cv2.COLOR_BGR2RGB) if draw else img_rgb
 
-    except Exception as e:
-        return {"Status": 1.0}, img_rgb, f"⚠ Inference failed: {e}"
+    status = f" Detected emotion: {pred_label}"
+    return confidence, img_out, status
 
 
 # ---- Gradio UI ----
@@ -117,7 +84,7 @@ with gr.Blocks(theme="default") as demo:
         with gr.Column(scale=1):
             image_input = gr.Image(
                 type="pil",
-                label="Upload Image or Use Webcam",
+                label="Examples",
                 sources=["upload", "webcam"],
                 interactive=True,
             )
@@ -127,16 +94,13 @@ with gr.Blocks(theme="default") as demo:
                 label="Draw Landmarks",
                 interactive=True,
             )
-            gr.Markdown(
-                "Tip: Switch **Draw Landmarks** ON to visualize key points (if your `util.get_face_landmarks` draws them)."
-            )
+            
 
         with gr.Column(scale=1):
             label_output = gr.Label(num_top_classes=3, label="Predicted Emotion & Confidence")
-            image_output = gr.Image(type="numpy", label="Annotated Image")
+            image_output = gr.Image(type="numpy", label="Image Output")
             status_output = gr.Textbox(label="Status", interactive=False)
 
-    # Examples (ensure these files exist in /examples)
     gr.Examples(
         examples=[
             ["examples/happy.png", "OFF"],
@@ -147,7 +111,7 @@ with gr.Blocks(theme="default") as demo:
         label="Try examples",
     )
 
-    # Real-time: changing either the image or the toggle re-runs inference automatically
+    # Real-time: change triggers inference
     image_input.change(
         fn=predict_emotion,
         inputs=[image_input, draw_toggle],