Update app.py

app.py

@@ -23,64 +23,67 @@ def load_model():
 
 model = load_model()
 
-# --- PAGE SELECTION BUTTONS (visible on mobile too) ---
+# --- LAYOUT ---
 st.title("🍽️ NutriDetect: AI Nutrition Companion")
-page = st.radio("Navigate to", ["Home", "Try It", "About"], horizontal=True)
-
-# --- HOME PAGE ---
-if page == "Home":
-    st.image("back_image.png", width=800)
-    st.markdown("""
-    Welcome to **NutriDetect** – detect food items from images using a custom-trained YOLOv8 model.
-
-    - 📷 Upload food images
-    - 🧠 Detect with YOLO
-    - ⚡ Built for mobile and real-time use
-    """)
-
-# --- TRY IT PAGE ---
-elif page == "Try It":
-    st.header("🚀 Try Food Detection")
-    uploaded_file = st.file_uploader("Upload a food image", type=["jpg", "jpeg", "png"])
-
-    if uploaded_file:
-        img_id = str(uuid.uuid4()) + ".jpg"
-        image_path = os.path.join(TEMP_IMAGE_DIR, img_id)
-        with open(image_path, "wb") as f:
-            f.write(uploaded_file.read())
-        st.image(image_path, caption="Uploaded Image", width=400)
-
-        if st.button("🔍 Detect Food"):
-            with st.spinner("Detecting..."):
-                results = model.predict(source=image_path, save=True, save_txt=False)
-
-                # Find latest prediction folder
-                detect_dir = "runs/detect"
-                all_preds = sorted(os.listdir(detect_dir), key=lambda x: os.path.getmtime(os.path.join(detect_dir, x)))
-                latest_run = os.path.join(detect_dir, all_preds[-1])
-                pred_image_path = os.path.join(latest_run, img_id)
-
-                if os.path.exists(pred_image_path):
-                    st.image(pred_image_path, caption="Detection Result", width=800)
-                else:
-                    st.warning("Prediction image not found.")
-
-                boxes = results[0].boxes
-                names = model.names
-                st.subheader("📋 Detected Items")
-                if len(boxes) == 0:
-                    st.warning("No objects detected.")
-                else:
-                    for box in boxes:
-                        class_id = int(box.cls)
-                        conf = float(box.conf)
-                        label = names[class_id]
-                        st.markdown(f"- **{label}** (Confidence: {conf:.2f})")
-
-# --- ABOUT PAGE ---
-elif page == "About":
-    st.header("📖 About NutriDetect")
-    st.markdown("""
+
+col_nav, col_main = st.columns([1, 5])  # Navigation on the left
+
+with col_nav:
+    st.markdown("### 🔎 Menu")
+    page = st.radio("", ["Home", "Try It", "About"])
+
+with col_main:
+    if page == "Home":
+        st.image("back_image.png", width=800)
+        st.markdown("""
+        Welcome to **NutriDetect** – detect food items from images using a custom-trained YOLOv8 model.
+
+        - 📷 Upload food images  
+        - 🧠 Detect with YOLO  
+        - ⚡ Built for mobile and real-time use
+        """)
+
+    elif page == "Try It":
+        st.header("🚀 Try Food Detection")
+        uploaded_file = st.file_uploader("Upload a food image", type=["jpg", "jpeg", "png"])
+
+        if uploaded_file:
+            img_id = str(uuid.uuid4()) + ".jpg"
+            image_path = os.path.join(TEMP_IMAGE_DIR, img_id)
+            with open(image_path, "wb") as f:
+                f.write(uploaded_file.read())
+            st.image(image_path, caption="Uploaded Image", width=400)
+
+            if st.button("🔍 Detect Food"):
+                with st.spinner("Detecting..."):
+                    results = model.predict(source=image_path, save=True, save_txt=False)
+
+                    # Find the latest result image
+                    detect_dir = "runs/detect"
+                    all_preds = sorted(os.listdir(detect_dir), key=lambda x: os.path.getmtime(os.path.join(detect_dir, x)))
+                    latest_run = os.path.join(detect_dir, all_preds[-1])
+                    pred_image_path = os.path.join(latest_run, img_id)
+
+                    if os.path.exists(pred_image_path):
+                        st.image(pred_image_path, caption="Detection Result", width=800)
+                    else:
+                        st.warning("Prediction image not found.")
+
+                    boxes = results[0].boxes
+                    names = model.names
+                    st.subheader("📋 Detected Items")
+                    if len(boxes) == 0:
+                        st.warning("No objects detected.")
+                    else:
+                        for box in boxes:
+                            class_id = int(box.cls)
+                            conf = float(box.conf)
+                            label = names[class_id]
+                            st.markdown(f"- **{label}** (Confidence: {conf:.2f})")
+
+    elif page == "About":
+        st.header("📖 About NutriDetect")
+        st.markdown("""
 **NutriDetect** is an AI-powered food recognition system developed to simplify dietary assessment and support healthier eating habits.
 
 This project was accepted at the **ACM Conference**, showcasing innovation in AI-powered nutritional informatics.
@@ -95,17 +98,17 @@ NutriDetect uses the YOLOv8 object detection model trained on over 55,000 curate
 
 ### 🧠 Why It Matters
 
-- Enables quick and accurate food recognition
-- Designed for mobile, AR, and wearable integration
+- Enables quick and accurate food recognition  
+- Designed for mobile, AR, and wearable integration  
 - Built with a culturally diverse dataset for global usage
 
 ---
 
 ### 🔬 Research Highlights
 
-- mAP@50 of **75.9%**
-- Precision of **72.4%**
-- Trained on NVIDIA RTX 4060 (8GB VRAM)
+- mAP@50 of **75.9%**  
+- Precision of **72.4%**  
+- Trained on NVIDIA RTX 4060 (8GB VRAM)  
 - Dataset curated from UECFOOD256, OID, and more
 
 *NutriDetect is a step toward smarter, AI-powered food logging and healthcare.*