Update app.py

app.py

@@ -1,364 +1,427 @@
+"""
+Indian/Pakistani Food Classifier
+A deep learning model to identify 80+ Indian and Pakistani dishes
+"""
+
 import streamlit as st
 import tensorflow as tf
 import numpy as np
 from PIL import Image
 import json
 import os
-
-# Define custom metrics (MUST be defined before loading model)
-def top_3_accuracy(y_true, y_pred):
-    return tf.keras.metrics.top_k_categorical_accuracy(y_true, y_pred, k=3)
-
-def top_5_accuracy(y_true, y_pred):
-    return tf.keras.metrics.top_k_categorical_accuracy(y_true, y_pred, k=5)
-
-# Register custom functions with Keras
-@tf.keras.utils.register_keras_serializable()
-def top_3_accuracy_serializable(y_true, y_pred):
-    return tf.keras.metrics.top_k_categorical_accuracy(y_true, y_pred, k=3)
-
-@tf.keras.utils.register_keras_serializable()
-def top_5_accuracy_serializable(y_true, y_pred):
-    return tf.keras.metrics.top_k_categorical_accuracy(y_true, y_pred, k=5)
-
-# Custom objects dictionary
-custom_objects = {
-    'top_3_accuracy': top_3_accuracy,
-    'top_5_accuracy': top_5_accuracy,
-    'top_3_accuracy_serializable': top_3_accuracy_serializable,
-    'top_5_accuracy_serializable': top_5_accuracy_serializable
-}
+import plotly.graph_objects as go
+import pandas as pd
+from datetime import datetime
+import random
 
 # Page configuration
 st.set_page_config(
-    page_title="Pakistan Food Classifier",
+    page_title="Pakistani & Indian Food Classifier",
     page_icon="🍛",
     layout="wide",
     initial_sidebar_state="expanded"
 )
 
-# Custom CSS for Pakistan theme
+# Custom CSS for beautiful UI
 st.markdown("""
-    <style>
-    .stApp {
-        background-color: #f5f5f5;
+<style>
+    /* Main container */
+    .main {
+        background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
     }
-    .main-header {
-        text-align: center;
+    
+    /* Header styling */
+    .header-container {
+        background: linear-gradient(135deg, #006400 0%, #008000 50%, #ffffff 100%);
         padding: 2rem;
-        background: linear-gradient(135deg, #006400 0%, #ffffff 50%, #006400 100%);
-        color: white;
-        border-radius: 10px;
+        border-radius: 20px;
         margin-bottom: 2rem;
+        text-align: center;
+        box-shadow: 0 10px 30px rgba(0,0,0,0.1);
+    }
+    
+    .header-title {
+        font-size: 3rem;
+        font-weight: bold;
+        color: white;
+        margin-bottom: 0.5rem;
+    }
+    
+    .header-subtitle {
+        font-size: 1.2rem;
+        color: rgba(255,255,255,0.9);
+    }
+    
+    .pakistan-flag {
+        font-size: 2rem;
+        margin-bottom: 1rem;
     }
+    
+    /* Card styling */
     .prediction-card {
-        background-color: white;
+        background: white;
+        border-radius: 15px;
         padding: 1.5rem;
-        border-radius: 10px;
-        box-shadow: 0 2px 4px rgba(0,0,0,0.1);
         margin: 1rem 0;
+        box-shadow: 0 4px 6px rgba(0,0,0,0.1);
+        transition: transform 0.3s;
     }
-    .top1 {
+    
+    .prediction-card:hover {
+        transform: translateY(-5px);
+        box-shadow: 0 6px 12px rgba(0,0,0,0.15);
+    }
+    
+    /* Top prediction styling */
+    .top-prediction {
         background: linear-gradient(135deg, #006400 0%, #008000 100%);
         color: white;
+        border-radius: 15px;
+        padding: 1.5rem;
+        margin: 1rem 0;
+        text-align: center;
+    }
+    
+    .top-prediction h2 {
+        font-size: 2.5rem;
+        margin: 0;
+    }
+    
+    .confidence-score {
+        font-size: 1.2rem;
+        margin-top: 0.5rem;
+    }
+    
+    /* Other predictions */
+    .other-prediction {
+        background: #f8f9fa;
+        border-left: 5px solid #006400;
+        border-radius: 10px;
         padding: 1rem;
+        margin: 0.8rem 0;
+    }
+    
+    /* Sidebar styling */
+    .sidebar-content {
+        background: #f0f2f6;
         border-radius: 10px;
-        margin: 0.5rem 0;
+        padding: 1rem;
     }
-    .top2 {
-        background-color: #e8f5e9;
-        padding: 0.8rem;
-        border-radius: 8px;
-        margin: 0.5rem 0;
-        border-left: 5px solid #006400;
+    
+    /* Button styling */
+    .stButton > button {
+        background: linear-gradient(135deg, #006400 0%, #008000 100%);
+        color: white;
+        border: none;
+        padding: 0.5rem 2rem;
+        border-radius: 25px;
+        font-weight: bold;
+        transition: all 0.3s;
     }
-    .top3 {
-        background-color: #f1f8e9;
-        padding: 0.8rem;
-        border-radius: 8px;
-        margin: 0.5rem 0;
-        border-left: 5px solid #8bc34a;
+    
+    .stButton > button:hover {
+        transform: scale(1.05);
+        box-shadow: 0 4px 8px rgba(0,0,0,0.2);
     }
-    .pakistan-flag {
+    
+    /* Footer */
+    .footer {
         text-align: center;
-        font-size: 1.2rem;
+        padding: 2rem;
+        color: #666;
+        font-size: 0.8rem;
+        margin-top: 3rem;
+    }
+    
+    /* Success/Error messages */
+    .success-message {
+        background: #d4edda;
+        color: #155724;
+        padding: 1rem;
+        border-radius: 10px;
         margin: 1rem 0;
     }
-    </style>
+    
+    .info-message {
+        background: #d1ecf1;
+        color: #0c5460;
+        padding: 1rem;
+        border-radius: 10px;
+        margin: 1rem 0;
+    }
+</style>
 """, unsafe_allow_html=True)
 
-# Pakistan flag emoji in header
-st.markdown('<div class="pakistan-flag">🇵🇰 Pakistan 🇵🇰</div>', unsafe_allow_html=True)
-
-# Paths for model files
-MODEL_PATH = "dish_classifier_final.keras"
-CLASSES_PATH = "class_names.json"
-
+# ============================================================
+# LOAD MODEL AND CLASSES
+# ============================================================
 @st.cache_resource
 def load_model():
-    """Load the trained model with custom objects"""
-    
-    # Check if model exists
-    if not os.path.exists(MODEL_PATH):
-        st.error(f"""
-        ❌ Model file not found: `{MODEL_PATH}`
-        
-        **Please upload the model file:**
-        1. Click on **"Files"** tab above
-        2. Click **"Add file"** → **"Upload file"**
-        3. Upload `dish_classifier_final.keras`
-        4. Upload `class_names.json`
-        5. Refresh this page
-        
-        **Note:** The model file should be around 80-100 MB.
-        """)
-        return None
-    
+    """Load the trained model"""
     try:
-        # Load model with custom objects
-        model = tf.keras.models.load_model(
-            MODEL_PATH,
-            custom_objects=custom_objects,
-            compile=False
-        )
-        
-        # Recompile the model with the same metrics
-        model.compile(
-            optimizer='adam',
-            loss='categorical_crossentropy',
-            metrics=['accuracy', top_3_accuracy, top_5_accuracy]
-        )
-        
-        st.success("✅ Model loaded successfully!")
+        model = tf.keras.models.load_model('indian_food_classifier.keras')
         return model
-    except Exception as e:
-        st.error(f"Error loading model: {str(e)}")
-        st.info("Make sure the model file is not corrupted and was trained with TensorFlow 2.x")
-        return None
+    except:
+        try:
+            model = tf.keras.models.load_model('/kaggle/working/indian_food_classifier.keras')
+            return model
+        except:
+            st.error("⚠️ Model file not found. Please upload 'indian_food_classifier.keras'")
+            return None
 
 @st.cache_data
 def load_class_names():
-    """Load class names from JSON file"""
-    if not os.path.exists(CLASSES_PATH):
-        st.error(f"❌ Class names file not found: `{CLASSES_PATH}`")
-        st.info("Please upload `class_names.json` file")
-        return None
-    
+    """Load class names"""
     try:
-        with open(CLASSES_PATH, 'r') as f:
+        with open('class_names.json', 'r') as f:
             class_names = json.load(f)
-        st.success(f"✅ Loaded {len(class_names)} Pakistani food classes!")
         return class_names
-    except Exception as e:
-        st.error(f"Error loading class names: {e}")
-        return None
+    except:
+        try:
+            with open('/kaggle/working/class_names.json', 'r') as f:
+                class_names = json.load(f)
+            return class_names
+        except:
+            st.error("⚠️ class_names.json not found. Please upload the file.")
+            return None
 
-def preprocess_image(image):
+def preprocess_image(image, target_size=(224, 224)):
     """Preprocess image for model prediction"""
-    # Convert to RGB if needed
     if image.mode != 'RGB':
         image = image.convert('RGB')
-    
-    # Resize to 224x224
-    image = image.resize((224, 224))
-    
-    # Convert to array
-    img_array = np.array(image)
-    
-    # Normalize to [0,1] range
-    img_array = img_array / 255.0
-    
-    # Expand dimensions to create batch
+    image = image.resize(target_size)
+    img_array = np.array(image) / 255.0
     img_array = np.expand_dims(img_array, axis=0)
-    
     return img_array
 
-def predict_image(model, class_names, image):
-    """Make prediction on uploaded image"""
-    processed_image = preprocess_image(image)
-    
-    # Get predictions
-    predictions = model.predict(processed_image, verbose=0)[0]
+def format_food_name(name):
+    """Format food name for display"""
+    return name.replace('_', ' ').title()
+
+def create_confidence_chart(confidences, labels, top_n=5):
+    """Create an interactive confidence chart"""
+    fig = go.Figure(data=[
+        go.Bar(
+            x=confidences[:top_n],
+            y=[format_food_name(l) for l in labels[:top_n]],
+            orientation='h',
+            marker=dict(
+                color=confidences[:top_n],
+                colorscale='Greens',
+                showscale=True,
+                colorbar=dict(title="Confidence (%)")
+            ),
+            text=[f"{c:.1f}%" for c in confidences[:top_n]],
+            textposition='outside'
+        )
+    ])
     
-    # Get top 5 predictions
-    top_5_idx = np.argsort(predictions)[-5:][::-1]
-    top_5_labels = [class_names[idx] for idx in top_5_idx]
-    top_5_probs = [predictions[idx] * 100 for idx in top_5_idx]
+    fig.update_layout(
+        title="Top Predictions Confidence Score",
+        xaxis_title="Confidence (%)",
+        yaxis_title="Food Item",
+        height=400,
+        margin=dict(l=0, r=0, t=40, b=0),
+        paper_bgcolor='rgba(0,0,0,0)',
+        plot_bgcolor='rgba(0,0,0,0)'
+    )
     
-    return top_5_labels, top_5_probs
+    return fig
 
+# ============================================================
+# MAIN APP
+# ============================================================
 def main():
-    # Header with Pakistan theme
+    # Header
     st.markdown("""
-    <div class="main-header">
-        <h1>🍛 Pakistan Food Classifier</h1>
-        <p>Upload a photo of Pakistani food and AI will identify it!</p>
-        <p style="font-size: 0.9rem; opacity: 0.9;">Supports 80+ Pakistani dishes with 85% Top-5 Accuracy</p>
+    <div class="header-container">
+        <div class="pakistan-flag">
+            🇵🇰 🇮🇳 🇵🇰
+        </div>
+        <div class="header-title">
+            🍛 Pakistani & Indian Food Classifier
+        </div>
+        <div class="header-subtitle">
+            AI-powered dish recognition for 80+ South Asian delicacies
+        </div>
     </div>
     """, unsafe_allow_html=True)
     
-    # Sidebar with Pakistan information
+    # Sidebar
     with st.sidebar:
-        st.markdown("### 🇵🇰 Model Information")
+        st.markdown("### 🏆 Model Information")
         st.info("""
-        - **Model:** EfficientNetV2S
-        - **Classes:** 80 Pakistani Dishes
-        - **Accuracy:** 56.25%
-        - **Top-3 Accuracy:** 77.38%
-        - **Top-5 Accuracy:** 84.62%
+        - **Architecture:** EfficientNetV2S
+        - **Classes:** 80 Indian/Pakistani Dishes
+        - **Accuracy:** 59.25%
+        - **Input Size:** 224x224 pixels
         """)
         
-        st.markdown("### 🍽️ Popular Pakistani Dishes")
-        st.markdown("""
-        - **Biryani** (Chicken, Mutton, Sindhi)
-        - **Nihari**
-        - **Haleem**
-        - **Karachi Haleem**
-        - **Chapli Kebab**
-        - **Seekh Kebab**
-        - **Karahi** (Chicken, Mutton)
-        - **Daal Chawal**
-        - **Naan** (Garlic, Plain, Roghni)
-        - **Halwa Puri**
-        - **Gulab Jamun**
-        - **Kheer**
-        - **And 68 more...**
-        """)
+        st.markdown("---")
+        st.markdown("### 🍽️ Popular Dishes")
+        
+        # Random popular dishes
+        popular_dishes = [
+            "Biryani", "Nihari", "Butter Chicken", "Aloo Gobi",
+            "Samosa", "Gulab Jamun", "Naan", "Haleem",
+            "Karahi", "Seekh Kebab", "Dal Makhani", "Ras Malai"
+        ]
         
-        st.markdown("### 📝 How to Use")
+        for dish in random.sample(popular_dishes, min(6, len(popular_dishes))):
+            st.markdown(f"• {dish}")
+        
+        st.markdown("---")
+        st.markdown("### 📊 How It Works")
         st.markdown("""
-        1. Click 'Browse files' below
-        2. Upload an image of Pakistani food
-        3. Wait for AI prediction
-        4. See top 5 predictions with confidence scores
+        1. 📸 Upload a clear photo of food
+        2. 🤖 AI analyzes the image
+        3. 🎯 Get top 5 predictions with confidence scores
+        4. 📈 View detailed confidence chart
         """)
         
         st.markdown("---")
-        st.markdown("### 📊 Model Performance")
-        
-        # Add simple gauges for accuracy
-        st.markdown("**Exact Match Accuracy**")
-        st.progress(0.56)
-        st.caption("56.25%")
-        
-        st.markdown("**Top-3 Accuracy**")
-        st.progress(0.77)
-        st.caption("77.38%")
-        
-        st.markdown("**Top-5 Accuracy**")
-        st.progress(0.85)
-        st.caption("84.62%")
+        st.markdown("### 💡 Tips for Best Results")
+        st.markdown("""
+        - Use well-lit photos
+        - Focus on the main dish
+        - Avoid cluttered backgrounds
+        - Single dish per photo works best
+        """)
         
         st.markdown("---")
-        st.markdown("Made with ❤️ for Pakistan")
+        st.markdown("Made with ❤️ for South Asian Cuisine")
     
     # Main content area
     col1, col2 = st.columns([1, 1])
     
     with col1:
-        st.markdown("### 📤 Upload Pakistani Food Image")
+        st.markdown("### 📤 Upload Food Image")
         uploaded_file = st.file_uploader(
             "Choose an image...",
             type=['jpg', 'jpeg', 'png', 'webp', 'gif'],
-            help="Upload a clear image of Pakistani food for best results"
+            help="Upload a clear image of Pakistani or Indian food"
         )
         
         if uploaded_file is not None:
-            # Display uploaded image
             image = Image.open(uploaded_file)
-            st.image(image, caption='📸 Uploaded Image', use_container_width=True)
             
-            # Add image info
-            st.caption(f"Image size: {image.size[0]} x {image.size[1]} pixels")
+            # Display image with styling
+            st.markdown("#### Preview")
+            st.image(image, caption="Uploaded Image", use_container_width=True)
+            
+            # Image info
+            st.caption(f"📐 Image size: {image.size[0]} x {image.size[1]} pixels")
     
     with col2:
         if uploaded_file is not None:
-            st.markdown("### 🔍 Prediction Results")
+            st.markdown("### 🔍 Analysis Results")
             
-            with st.spinner('🍛 Analyzing your Pakistani food image...'):
-                # Load model and class names
+            with st.spinner("🍛 Analyzing your food image..."):
+                # Load model and classes
                 model = load_model()
                 class_names = load_class_names()
                 
                 if model is not None and class_names is not None:
-                    # Make prediction
-                    top_5_labels, top_5_probs = predict_image(model, class_names, image)
+                    # Preprocess and predict
+                    processed_img = preprocess_image(image)
+                    predictions = model.predict(processed_img, verbose=0)[0]
                     
-                    # Display results in a card
-                    st.markdown('<div class="prediction-card">', unsafe_allow_html=True)
+                    # Get top 5 predictions
+                    top_5_idx = np.argsort(predictions)[-5:][::-1]
+                    top_5_names = [class_names[idx] for idx in top_5_idx]
+                    top_5_confidences = [predictions[idx] * 100 for idx in top_5_idx]
                     
-                    # Top 1 prediction (highlighted)
+                    # Display top prediction (highlighted)
                     st.markdown(f"""
-                    <div class="top1">
-                        <h3>🏆 Top Prediction</h3>
-                        <h2>🍽️ {top_5_labels[0]}</h2>
-                        <p>Confidence: {top_5_probs[0]:.2f}%</p>
+                    <div class="top-prediction">
+                        <div style="font-size: 1.2rem;">🥇 Top Prediction</div>
+                        <h2>{format_food_name(top_5_names[0])}</h2>
+                        <div class="confidence-score">Confidence: {top_5_confidences[0]:.2f}%</div>
                     </div>
                     """, unsafe_allow_html=True)
                     
-                    # Top 2 prediction
-                    st.markdown(f"""
-                    <div class="top2">
-                        <strong>🥈 Most Likely:</strong><br>
-                        <span style="font-size: 1.1rem;">{top_5_labels[1]}</span><br>
-                        <span style="color: #666;">Confidence: {top_5_probs[1]:.2f}%</span>
-                    </div>
-                    """, unsafe_allow_html=True)
+                    # Display other predictions
+                    st.markdown("#### Other Possibilities")
                     
-                    # Top 3 prediction
-                    st.markdown(f"""
-                    <div class="top3">
-                        <strong>🥉 Could Be:</strong><br>
-                        <span style="font-size: 1.1rem;">{top_5_labels[2]}</span><br>
-                        <span style="color: #666;">Confidence: {top_5_probs[2]:.2f}%</span>
-                    </div>
-                    """, unsafe_allow_html=True)
+                    for i in range(1, min(5, len(top_5_names))):
+                        confidence_percent = top_5_confidences[i]
+                        
+                        # Determine emoji based on rank
+                        if i == 1:
+                            emoji = "🥈"
+                        elif i == 2:
+                            emoji = "🥉"
+                        else:
+                            emoji = f"{i+1}️⃣"
+                        
+                        st.markdown(f"""
+                        <div class="other-prediction">
+                            <strong>{emoji} {format_food_name(top_5_names[i])}</strong><br>
+                            <span style="color: #666;">Confidence: {confidence_percent:.2f}%</span>
+                        </div>
+                        """, unsafe_allow_html=True)
                     
-                    # Top 4 & 5
-                    st.markdown(f"""
-                    <div style="margin-top: 1rem; padding: 0.5rem;">
-                        <p><strong>4️⃣ Other possibilities:</strong> {top_5_labels[3]} ({top_5_probs[3]:.1f}%)</p>
-                        <p><strong>5️⃣ Also possible:</strong> {top_5_labels[4]} ({top_5_probs[4]:.1f}%)</p>
-                    </div>
-                    """, unsafe_allow_html=True)
+                    # Confidence chart
+                    st.markdown("---")
+                    st.markdown("### 📊 Confidence Analysis")
                     
-                    st.markdown('</div>', unsafe_allow_html=True)
-                    
-                    # Add confidence visualization
-                    st.markdown("### 📊 Confidence Level")
-                    if top_5_probs[0] > 70:
-                        confidence_color = "🟢 High Confidence"
-                    elif top_5_probs[0] > 50:
-                        confidence_color = "🟡 Medium Confidence"
-                    else:
-                        confidence_color = "🟠 Low Confidence"
+                    fig = create_confidence_chart(top_5_confidences, top_5_names, top_n=5)
+                    st.plotly_chart(fig, use_container_width=True)
                     
-                    st.markdown(f"{confidence_color} - {top_5_probs[0]:.1f}% confident this is **{top_5_labels[0]}**")
+                    # Confidence meter for top prediction
+                    st.markdown("#### Confidence Meter")
+                    confidence_level = top_5_confidences[0]
                     
-                    # Add suggestion for low confidence
-                    if top_5_probs[0] < 50:
-                        st.info("💡 **Tip:** Try uploading a clearer image of just the dish for better accuracy")
+                    if confidence_level > 70:
+                        st.success(f"🎯 High confidence! The AI is very sure this is {format_food_name(top_5_names[0])}")
+                    elif confidence_level > 50:
+                        st.warning(f"🤔 Medium confidence. The AI thinks it's {format_food_name(top_5_names[0])}")
+                    else:
+                        st.info(f"💡 Low confidence. Try uploading a clearer photo for better results")
     
-    # Add information when no image is uploaded
-    if not uploaded_file:
-        st.markdown("---")
-        st.markdown("### 🕌 About This Model")
-        
-        col1, col2, col3 = st.columns(3)
-        with col1:
-            st.info("**🍛 Training Data**\n\nTrained on 3,200 images of 80+ Pakistani dishes from Pakistan's diverse cuisine")
-        with col2:
-            st.info("**🤖 Model Architecture**\n\nUses EfficientNetV2S, a state-of-the-art CNN pre-trained on ImageNet")
-        with col3:
-            st.info("**🎯 Purpose**\n\nHelp food enthusiasts, restaurants, and researchers identify Pakistani dishes")
+    # Footer with additional information
+    st.markdown("---")
+    
+    col1, col2, col3 = st.columns(3)
+    
+    with col1:
+        st.markdown("""
+        ### 🎯 Supported Cuisines
+        - Punjabi
+        - Mughlai
+        - South Indian
+        - Sindhi
+        - Kashmiri
+        - Hyderabadi
+        """)
+    
+    with col2:
+        st.markdown("""
+        ### 🍜 Dish Categories
+        - Curries & Gravies
+        - Rice Dishes (Biryani)
+        - Breads (Naan, Roti)
+        - Desserts & Sweets
+        - Snacks & Appetizers
+        - Beverages
+        """)
+    
+    with col3:
+        st.markdown("""
+        ### 📈 Model Performance
+        - 59.25% Top-1 Accuracy
+        - 80+ Food Classes
+        - 3,200 Training Images
+        - EfficientNetV2S Backbone
+        - Real-time Predictions
+        """)
     
     # Footer
-    st.markdown("---")
     st.markdown("""
-    <div style="text-align: center; color: #666; font-size: 0.8rem;">
-        <p>🇵🇰 Celebrating Pakistan's rich culinary heritage 🇵🇰</p>
-        <p>⚠️ Note: Model accuracy is 56% for exact match, 85% for Top-5 predictions.<br>
-        For best results, use clear, well-lit photos of individual dishes.</p>
+    <div class="footer">
+        <p>🇵🇰 Celebrating the rich culinary heritage of Pakistan and India 🇮🇳</p>
+        <p>⚠️ Note: For best results, use clear, well-lit images of individual dishes. The model works best on traditional South Asian cuisine.</p>
+        <p>Made with Streamlit & TensorFlow | Model trained on 80+ dishes</p>
     </div>
     """, unsafe_allow_html=True)