app.py

import streamlit as st
import requests
import google.generativeai as genai
import firebase_admin
from firebase_admin import credentials, db
from PIL import Image
import numpy as np
import base64
from io import BytesIO
from tensorflow.keras.applications import MobileNetV2
from tensorflow.keras.applications.mobilenet_v2 import decode_predictions, preprocess_input

# Initialize Firebase
if not firebase_admin._apps:
    cred = credentials.Certificate("firebase_credentials.json")  
    firebase_admin.initialize_app(cred, {
        'databaseURL': 'https://binsight-beda0-default-rtdb.asia-southeast1.firebasedatabase.app/'
    })

# Configure Gemini 2.0 Flash (gemini-1.5-flash)
genai.configure(api_key="AIzaSyBREh8Uei7uDCbzPaYW2WdalOdjVWcQLAM")  # Replace with actual API key

# Load MobileNetV2 pre-trained model
mobilenet_model = MobileNetV2(weights="imagenet")

# Function to classify image
def classify_image(image):
    img = image.resize((224, 224))
    img_array = np.array(img)
    img_array = np.expand_dims(img_array, axis=0)
    img_array = preprocess_input(img_array)
    predictions = mobilenet_model.predict(img_array)
    labels = decode_predictions(predictions, top=5)[0]
    return {label[1]: round(float(label[2]) * 100, 2) for label in labels}

# Function to check if image is a dustbin
def is_dustbin_image(classification_results):
    dustbin_keywords = ["trash", "bin", "garbage", "waste", "dustbin", "ashcan", "recycle", "rubbish"]
    return any(any(keyword in label.lower() for keyword in dustbin_keywords) for label in classification_results.keys())

# Function to convert image to Base64
def convert_image_to_base64(image):
    buffered = BytesIO()
    image.save(buffered, format="PNG")
    return base64.b64encode(buffered.getvalue()).decode()

# Function to generate AI recommendations using Gemini 2.0 Flash
def get_genai_response(classification_results, location):
    try:
        classification_summary = "\n".join([f"- **{label}:** {score}%" for label, score in classification_results.items()])
        location_summary = f"""
        - **Latitude:** {location[0] if location[0] else 'N/A'}
        - **Longitude:** {location[1] if location[1] else 'N/A'}
        - **Address:** {location[2] if location[2] else 'N/A'}
        """
        
        prompt = f"""
        ### You are an environmental expert. Analyze the following waste classification:
        
        **1. Image Classification Results:**
        {classification_summary}
        
        **2. Location Details:**
        {location_summary}
        ### Required Analysis:
        - Describe the waste detected in the image.
        - Potential health & environmental risks.
        - Recommended disposal methods & precautions.
        - Eco-friendly alternatives.
        """

        model = genai.GenerativeModel("gemini-1.5-flash")
        response = model.generate_content(prompt)
        return response.text if response else "⚠️ No response received."
    except Exception as e:
        st.error(f"⚠️ Error using Generative AI: {e}")
        return None

# Extract query parameters
query_params = st.experimental_get_query_params()
latitude = query_params.get("lat", [""])[0]
longitude = query_params.get("lon", [""])[0]
address = query_params.get("addr", [""])[0]

# Display detected location
st.header("📍 Detected Location")
st.success(f"**Latitude:** {latitude}")
st.success(f"**Longitude:** {longitude}")
st.success(f"**Address:** {address}")

# Streamlit UI
st.title("🗑️ BinSight: Upload Dustbin Image")

uploaded_file = st.file_uploader("📷 Upload a dustbin image", type=["jpg", "jpeg", "png"])
if uploaded_file:
    image = Image.open(uploaded_file)
    st.image(image, caption="📸 Uploaded Image", use_column_width=True)
    
    image_base64 = convert_image_to_base64(image)
    classification_results = classify_image(image)
    
    st.header("🧪 Classification Results")
    for label, confidence in classification_results.items():
        st.write(f"✅ **{label}:** {confidence}%")
    
    if not is_dustbin_image(classification_results):
        st.error("⚠️ Please upload a valid dustbin image. No dustbin detected in the image.")
        st.stop()

    st.header("🧠 AI Analysis & Recommendations")
    gemini_response = get_genai_response(classification_results, (latitude, longitude, address))
    st.write(gemini_response)
    
    dustbin_data = {
        "latitude": latitude,
        "longitude": longitude,
        "address": address,
        "classification": classification_results,
        "allocated_truck": None,
        "status": "Pending",
        "image": image_base64
    }
    
    db.reference("dustbins").push(dustbin_data)
    st.success("✅ Dustbin data uploaded successfully!")
    
    # Thank You Message with Styling
    st.markdown(
        """
        <div style="
            background-color: #DFF2BF;
            color: #4F8A10;
            padding: 20px;
            border-radius: 10px;
            text-align: center;
            font-size: 18px;
            margin-top: 20px;">
            🎉 <b>Thank You for Your Contribution!</b> 🎉<br><br>
            Your effort in uploading this image helps keep our environment clean and green. 🌱♻️<br>
            Together, we make a difference! 🙌
        </div>
        """, unsafe_allow_html=True
    )

# Back button to redirect to dashboard
# st.markdown("<br>", unsafe_allow_html=True)
# st.markdown("<a href='https://binsight.onrender.com/dashboard.html' target='_self' style='text-decoration:none;'><button style='padding: 10px 20px; font-size: 16px;'>⬅ Back to Dashboard</button></a>", unsafe_allow_html=True)










# Best version without back button


# import streamlit as st
# import requests
# import google.generativeai as genai
# import firebase_admin
# from firebase_admin import credentials, db
# from PIL import Image
# import numpy as np
# import base64
# from io import BytesIO
# from tensorflow.keras.applications import MobileNetV2
# from tensorflow.keras.applications.mobilenet_v2 import decode_predictions, preprocess_input

# # Initialize Firebase
# if not firebase_admin._apps:
#     cred = credentials.Certificate("firebase_credentials.json")  
#     firebase_admin.initialize_app(cred, {
#         'databaseURL': 'https://binsight-beda0-default-rtdb.asia-southeast1.firebasedatabase.app/'
#     })

# # Configure Gemini 2.0 Flash (gemini-1.5-flash)
# genai.configure(api_key="AIzaSyBREh8Uei7uDCbzPaYW2WdalOdjVWcQLAM")  # Replace with actual API key

# # Load MobileNetV2 pre-trained model
# mobilenet_model = MobileNetV2(weights="imagenet")

# # Function to classify image
# def classify_image(image):
#     img = image.resize((224, 224))
#     img_array = np.array(img)
#     img_array = np.expand_dims(img_array, axis=0)
#     img_array = preprocess_input(img_array)
#     predictions = mobilenet_model.predict(img_array)
#     labels = decode_predictions(predictions, top=5)[0]
#     return {label[1]: round(float(label[2]) * 100, 2) for label in labels}

# # Function to check if image is a dustbin
# def is_dustbin_image(classification_results):
#     dustbin_keywords = ["trash", "bin", "garbage", "waste", "dustbin", "ashcan", "recycle", "rubbish"]
#     return any(any(keyword in label.lower() for keyword in dustbin_keywords) for label in classification_results.keys())

# # Function to convert image to Base64
# def convert_image_to_base64(image):
#     buffered = BytesIO()
#     image.save(buffered, format="PNG")
#     return base64.b64encode(buffered.getvalue()).decode()

# # Function to generate AI recommendations using Gemini 2.0 Flash
# def get_genai_response(classification_results, location):
#     try:
#         classification_summary = "\n".join([f"- **{label}:** {score}%" for label, score in classification_results.items()])
#         location_summary = f"""
#         - **Latitude:** {location[0] if location[0] else 'N/A'}
#         - **Longitude:** {location[1] if location[1] else 'N/A'}
#         - **Address:** {location[2] if location[2] else 'N/A'}
#         """
        
#         prompt = f"""
#         ### You are an environmental expert. Analyze the following waste classification:
        
#         **1. Image Classification Results:**
#         {classification_summary}
        
#         **2. Location Details:**
#         {location_summary}

#         ### Required Analysis:
#         - Describe the waste detected in the image.
#         - Potential health & environmental risks.
#         - Recommended disposal methods & precautions.
#         - Eco-friendly alternatives.
#         """

#         model = genai.GenerativeModel("gemini-1.5-flash")
#         response = model.generate_content(prompt)
#         return response.text if response else "⚠️ No response received."
#     except Exception as e:
#         st.error(f"⚠️ Error using Generative AI: {e}")
#         return None

# # Extract query parameters
# query_params = st.experimental_get_query_params()
# latitude = query_params.get("lat", [""])[0]
# longitude = query_params.get("lon", [""])[0]
# address = query_params.get("addr", [""])[0]

# # Display detected location
# st.header("📍 Detected Location")
# st.success(f"**Latitude:** {latitude}")
# st.success(f"**Longitude:** {longitude}")
# st.success(f"**Address:** {address}")

# # Streamlit UI
# st.title("🗑️ BinSight: Upload Dustbin Image")

# uploaded_file = st.file_uploader("📷 Upload a dustbin image", type=["jpg", "jpeg", "png"])
# if uploaded_file:
#     image = Image.open(uploaded_file)
#     st.image(image, caption="📸 Uploaded Image", use_column_width=True)
    
#     image_base64 = convert_image_to_base64(image)
#     classification_results = classify_image(image)
    
#     st.header("🧪 Classification Results")
#     for label, confidence in classification_results.items():
#         st.write(f"✅ **{label}:** {confidence}%")
    
#     if not is_dustbin_image(classification_results):
#         st.error("⚠️ Please upload a valid dustbin image. No dustbin detected in the image.")
#         st.stop()

#     st.header("🧠 AI Analysis & Recommendations")
#     gemini_response = get_genai_response(classification_results, (latitude, longitude, address))
#     st.write(gemini_response)
    
#     dustbin_data = {
#         "latitude": latitude,
#         "longitude": longitude,
#         "address": address,
#         "classification": classification_results,
#         "allocated_truck": None,
#         "status": "Pending",
#         "image": image_base64
#     }
    
#     db.reference("dustbins").push(dustbin_data)
#     st.success("✅ Dustbin data uploaded successfully!")
    
#     # Thank You Message with Styling
#     st.markdown(
#         """
#         <div style="
#             background-color: #DFF2BF;
#             color: #4F8A10;
#             padding: 20px;
#             border-radius: 10px;
#             text-align: center;
#             font-size: 18px;
#             margin-top: 20px;">
#             🎉 <b>Thank You for Your Contribution!</b> 🎉<br><br>
#             Your effort in uploading this image helps keep our environment clean and green. 🌱♻️<br>
#             Together, we make a difference! 🙌
#         </div>
#         """, unsafe_allow_html=True
#     )











# BEST version till now
# import streamlit as st
# import requests
# import google.generativeai as genai
# import firebase_admin
# from firebase_admin import credentials, db
# from PIL import Image
# import numpy as np
# import base64
# from io import BytesIO
# from tensorflow.keras.applications import MobileNetV2
# from tensorflow.keras.applications.mobilenet_v2 import decode_predictions, preprocess_input

# # Initialize Firebase
# if not firebase_admin._apps:
#     cred = credentials.Certificate("firebase_credentials.json")  
#     firebase_admin.initialize_app(cred, {
#         'databaseURL': 'https://binsight-beda0-default-rtdb.asia-southeast1.firebasedatabase.app/'
#     })

# # Configure Gemini 2.0 Flash (gemini-1.5-flash)
# genai.configure(api_key="AIzaSyBREh8Uei7uDCbzPaYW2WdalOdjVWcQLAM")  # Replace with actual API key

# # Load MobileNetV2 pre-trained model
# mobilenet_model = MobileNetV2(weights="imagenet")

# # Function to classify image
# def classify_image(image):
#     img = image.resize((224, 224))
#     img_array = np.array(img)
#     img_array = np.expand_dims(img_array, axis=0)
#     img_array = preprocess_input(img_array)
#     predictions = mobilenet_model.predict(img_array)
#     labels = decode_predictions(predictions, top=5)[0]
#     return {label[1]: round(float(label[2]) * 100, 2) for label in labels}  # Convert confidence to percentage

# # Function to check if image is a dustbin
# def is_dustbin_image(classification_results):
#     dustbin_keywords = ["trash", "bin", "garbage", "waste", "dustbin", "ashcan", "recycle", "rubbish"]
#     for label, confidence in classification_results.items():
#         if any(keyword in label.lower() for keyword in dustbin_keywords):
#             return True
#     return False

# # Function to convert image to Base64
# def convert_image_to_base64(image):
#     buffered = BytesIO()
#     image.save(buffered, format="PNG")
#     return base64.b64encode(buffered.getvalue()).decode()

# # Function to generate AI recommendations using Gemini 2.0 Flash
# def get_genai_response(classification_results, location):
#     try:
#         classification_summary = "\n".join([f"- **{label}:** {score}%" for label, score in classification_results.items()])
#         location_summary = f"""
#         - **Latitude:** {location[0] if location[0] else 'N/A'}
#         - **Longitude:** {location[1] if location[1] else 'N/A'}
#         - **Address:** {location[2] if location[2] else 'N/A'}
#         """

#         prompt = f"""
#         ### You are an environmental expert. Analyze the following waste classification:
        
#         **1. Image Classification Results:**
#         {classification_summary}
        
#         **2. Location Details:**
#         {location_summary}

#         ### Required Analysis:
#         - Describe the waste detected in the image.
#         - Potential health & environmental risks.
#         - Recommended disposal methods & precautions.
#         - Eco-friendly alternatives.
#         """

#         model = genai.GenerativeModel("gemini-1.5-flash")  # Using Gemini 2.0 Flash
#         response = model.generate_content(prompt)
#         return response.text if response else "⚠️ No response received."
#     except Exception as e:
#         st.error(f"⚠️ Error using Generative AI: {e}")
#         return None

# # **Fix: Revert to `st.experimental_get_query_params()` for full location display**
# query_params = st.experimental_get_query_params()

# latitude = query_params.get("lat", [""])[0]  # Extract full latitude
# longitude = query_params.get("lon", [""])[0]  # Extract full longitude
# address = query_params.get("addr", [""])[0]  # Extract full address

# # **Ensure full location values are displayed correctly**
# st.header("📍 Detected Location")
# st.success(f"**Latitude:** {latitude}")
# st.success(f"**Longitude:** {longitude}")
# st.success(f"**Address:** {address}")

# # Streamlit App UI
# st.title("🗑️ BinSight: Upload Dustbin Image")

# uploaded_file = st.file_uploader("📷 Upload a dustbin image", type=["jpg", "jpeg", "png"])
# if uploaded_file:
#     image = Image.open(uploaded_file)
#     st.image(image, caption="📸 Uploaded Image", use_container_width=True)

#     # Convert image to Base64
#     image_base64 = convert_image_to_base64(image)

#     # Classify Image
#     classification_results = classify_image(image)

#     # Display classification results
#     st.header("🧪 Classification Results")
#     if classification_results:
#         for label, confidence in classification_results.items():
#             st.write(f"✅ **{label}:** {confidence}%")
#     else:
#         st.error("⚠️ No classification results found.")
#         st.stop()

#     # **NEW CONDITION**: Ensure image is a dustbin before proceeding
#     if not is_dustbin_image(classification_results):
#         st.error("⚠️ Please upload a valid dustbin image. No dustbin detected in the image.")
#         st.stop()

#     # **Generate AI insights (Only Display, Not Store)**
#     st.header("🧠 AI Analysis & Recommendations")
#     gemini_response = get_genai_response(classification_results, (latitude, longitude, address))
#     st.write(gemini_response)

#     # Save only location, classification, and image in Firebase
#     dustbin_data = {
#         "latitude": latitude,
#         "longitude": longitude,
#         "address": address,
#         "classification": classification_results,
#         "allocated_truck": None,
#         "status": "Pending",
#         "image": image_base64
#     }

#     db.reference("dustbins").push(dustbin_data)  # Save data to Firebase
#     st.success("✅ Dustbin data uploaded successfully!")













## working well for embedded in html page but it is not taking full location

#import streamlit as st
# import requests
# import google.generativeai as genai
# import firebase_admin
# from firebase_admin import credentials, db
# from PIL import Image
# import numpy as np
# import base64
# from io import BytesIO
# from tensorflow.keras.applications import MobileNetV2
# from tensorflow.keras.applications.mobilenet_v2 import decode_predictions, preprocess_input

# # Initialize Firebase
# if not firebase_admin._apps:
#     cred = credentials.Certificate("firebase_credentials.json")  
#     firebase_admin.initialize_app(cred, {
#         'databaseURL': 'https://binsight-beda0-default-rtdb.asia-southeast1.firebasedatabase.app/'
#     })

# # Configure Gemini AI
# genai.configure(api_key="AIzaSyBREh8Uei7uDCbzPaYW2WdalOdjVWcQLAM")  # Replace with your actual API key

# # Load MobileNetV2 pre-trained model
# mobilenet_model = MobileNetV2(weights="imagenet")

# # Function to classify image
# def classify_image(image):
#     img = image.resize((224, 224))
#     img_array = np.array(img)
#     img_array = np.expand_dims(img_array, axis=0)
#     img_array = preprocess_input(img_array)
#     predictions = mobilenet_model.predict(img_array)
#     labels = decode_predictions(predictions, top=5)[0]
#     return {label[1]: round(float(label[2]) * 100, 2) for label in labels}  # Convert confidence to percentage

# # Function to convert image to Base64
# def convert_image_to_base64(image):
#     buffered = BytesIO()
#     image.save(buffered, format="PNG")
#     return base64.b64encode(buffered.getvalue()).decode()

# # Function to interact with Gemini AI (Only Show Results, Do NOT Save to Firebase)
# def get_genai_response(classification_results, location):
#     try:
#         classification_summary = "\n".join([f"- **{label}:** {score}%" for label, score in classification_results.items()])
#         location_summary = f"""
#         - **Latitude:** {location[0] if location[0] else 'N/A'}
#         - **Longitude:** {location[1] if location[1] else 'N/A'}
#         - **Address:** {location[2] if location[2] else 'N/A'}
#         """

#         prompt = f"""
#         ### You are an environmental expert. Analyze the following waste classification:
        
#         **1. Image Classification Results:**
#         {classification_summary}
        
#         **2. Location Details:**
#         {location_summary}

#         ### Required Analysis:
#         - Describe the waste detected in the image.
#         - Potential health & environmental risks.
#         - Recommended disposal methods & precautions.
#         - Eco-friendly alternatives.
#         """

#         model = genai.GenerativeModel("gemini-1.5-flash")
#         response = model.generate_content(prompt)
#         return response.text if response else "⚠️ No response received."
#     except Exception as e:
#         st.error(f"⚠️ Error using Generative AI: {e}")
#         return None

# # Extract location data from URL parameters
# query_params = st.query_params
# latitude = str(query_params.get("lat", [None])[0])  # Convert to string
# longitude = str(query_params.get("lon", [None])[0])  # Convert to string
# address = str(query_params.get("addr", [""])[0])  # Convert to full address

# # Display user location
# st.header("📍 Detected Location")
# if latitude and longitude:
#     st.success(f"**Latitude:** {latitude}, **Longitude:** {longitude}")
#     st.success(f"**Address:** {address}")
# else:
#     st.error("⚠️ Location data not received. Enable location detection in the main page.")

# # Streamlit App UI
# st.title("🗑️ BinSight: Upload Dustbin Image")

# uploaded_file = st.file_uploader("📷 Upload a dustbin image", type=["jpg", "jpeg", "png"])
# if uploaded_file:
#     image = Image.open(uploaded_file)
#     st.image(image, caption="📸 Uploaded Image", use_container_width=True)

#     # Convert image to Base64
#     image_base64 = convert_image_to_base64(image)

#     # Classify Image
#     classification_results = classify_image(image)

#     # Display classification results
#     st.header("🧪 Classification Results")
#     if classification_results:
#         for label, confidence in classification_results.items():
#             st.write(f"✅ **{label}:** {confidence}%")
#     else:
#         st.error("⚠️ No classification results found.")

#     # Get AI insights but DO NOT save to Firebase
#     st.header("🧠 AI Analysis & Recommendations")
#     gemini_response = get_genai_response(classification_results, (latitude, longitude, address))
#     st.write(gemini_response)

#     # Save only location, classification, and image in Firebase
#     dustbin_data = {
#         "latitude": latitude,
#         "longitude": longitude,
#         "address": address,
#         "classification": classification_results,
#         "allocated_truck": None,
#         "status": "Pending",
#         "image": image_base64
#     }

#     db.reference("dustbins").push(dustbin_data)  # Save data to Firebase
#     st.success("✅ Dustbin data uploaded successfully!")













## Best but not working location proper

# import streamlit as st
# import requests
# import firebase_admin
# from firebase_admin import credentials, db, auth
# from PIL import Image
# import numpy as np
# from geopy.geocoders import Nominatim
# from tensorflow.keras.applications import MobileNetV2
# from tensorflow.keras.applications.mobilenet_v2 import decode_predictions, preprocess_input
# import json
# import base64
# from io import BytesIO

# # Initialize Firebase (Check if already initialized)
# if not firebase_admin._apps:
#     cred = credentials.Certificate("firebase_credentials.json")  
#     firebase_admin.initialize_app(cred, {
#         'databaseURL': 'https://binsight-beda0-default-rtdb.asia-southeast1.firebasedatabase.app/'
#     })

# # Load MobileNetV2 pre-trained model
# mobilenet_model = MobileNetV2(weights="imagenet")

# # Function to classify the uploaded image using MobileNetV2
# def classify_image_with_mobilenet(image):
#     try:
#         img = image.resize((224, 224))
#         img_array = np.array(img)
#         img_array = np.expand_dims(img_array, axis=0)
#         img_array = preprocess_input(img_array)
#         predictions = mobilenet_model.predict(img_array)
#         labels = decode_predictions(predictions, top=5)[0]
#         return {label[1]: float(label[2]) for label in labels}
#     except Exception as e:
#         st.error(f"Error during image classification: {e}")
#         return {}

# # Function to get user's location
# def get_user_location():
#     st.write("Fetching location, please allow location access in your browser.")
#     geolocator = Nominatim(user_agent="binsight")
#     try:
#         ip_info = requests.get("https://ipinfo.io/json").json()
#         loc = ip_info.get("loc", "").split(",")
#         latitude, longitude = loc[0], loc[1] if len(loc) == 2 else (None, None)
#         if latitude and longitude:
#             address = geolocator.reverse(f"{latitude}, {longitude}").address
#             return latitude, longitude, address
#     except Exception as e:
#         st.error(f"Error retrieving location: {e}")
#     return None, None, None

# # Function to convert image to Base64
# def convert_image_to_base64(image):
#     buffered = BytesIO()
#     image.save(buffered, format="PNG")  # Convert to PNG format
#     img_str = base64.b64encode(buffered.getvalue()).decode()  # Encode as Base64
#     return img_str

# # User Login
# st.sidebar.header("User Login")
# user_email = st.sidebar.text_input("Enter your email")
# login_button = st.sidebar.button("Login")

# if login_button:
#     if user_email:
#         st.session_state["user_email"] = user_email
#         st.sidebar.success(f"Logged in as {user_email}")

# if "user_email" not in st.session_state:
#     st.warning("Please log in first.")
#     st.stop()

# # Get user location
# latitude, longitude, address = get_user_location()
# if latitude and longitude:
#     st.success(f"Location detected: {address}")
# else:
#     st.warning("Unable to fetch location, please enable location access.")
#     st.stop()

# # Streamlit App
# st.title("BinSight: Upload Dustbin Image")

# uploaded_file = st.file_uploader("Upload an image of the dustbin", type=["jpg", "jpeg", "png"])
# submit_button = st.button("Analyze and Upload")

# if submit_button and uploaded_file:
#     image = Image.open(uploaded_file)
#     st.image(image, caption="Uploaded Image", use_container_width=True)

#     # Convert image to Base64
#     image_base64 = convert_image_to_base64(image)

#     # Classify Image
#     classification_results = classify_image_with_mobilenet(image)
    
#     if classification_results:
#         db_ref = db.reference("dustbins")
#         dustbin_data = {
#             "user_email": st.session_state["user_email"],
#             "latitude": latitude,
#             "longitude": longitude,
#             "address": address,
#             "classification": classification_results,
#             "allocated_truck": None,
#             "status": "Pending",
#             "image": image_base64  # Store image as Base64 string
#         }
#         db_ref.push(dustbin_data)
#         st.success("Dustbin data uploaded successfully!")
#         st.write(f"**Location:** {address}")
#         st.write(f"**Latitude:** {latitude}, **Longitude:** {longitude}")
#     else:
#         st.error("Missing classification details. Cannot upload.")








# best without image

# import streamlit as st
# import requests
# import firebase_admin
# from firebase_admin import credentials, db, auth
# from PIL import Image
# import numpy as np
# from geopy.geocoders import Nominatim
# from tensorflow.keras.applications import MobileNetV2
# from tensorflow.keras.applications.mobilenet_v2 import decode_predictions, preprocess_input
# import json

# # Initialize Firebase
# if not firebase_admin._apps:
#     cred = credentials.Certificate("firebase_credentials.json")  
#     firebase_admin.initialize_app(cred, {
#         'databaseURL': 'https://binsight-beda0-default-rtdb.asia-southeast1.firebasedatabase.app/'
#     })

# # Load MobileNetV2 pre-trained model
# mobilenet_model = MobileNetV2(weights="imagenet")

# # Function to classify the uploaded image using MobileNetV2
# def classify_image_with_mobilenet(image):
#     try:
#         img = image.resize((224, 224))
#         img_array = np.array(img)
#         img_array = np.expand_dims(img_array, axis=0)
#         img_array = preprocess_input(img_array)
#         predictions = mobilenet_model.predict(img_array)
#         labels = decode_predictions(predictions, top=5)[0]
#         return {label[1]: float(label[2]) for label in labels}
#     except Exception as e:
#         st.error(f"Error during image classification: {e}")
#         return {}

# # Function to get user's location using geolocation API
# def get_user_location():
#     st.write("Fetching location, please allow location access in your browser.")
#     geolocator = Nominatim(user_agent="binsight")
#     try:
#         ip_info = requests.get("https://ipinfo.io/json").json()
#         loc = ip_info.get("loc", "").split(",")
#         latitude, longitude = loc[0], loc[1] if len(loc) == 2 else (None, None)
#         if latitude and longitude:
#             address = geolocator.reverse(f"{latitude}, {longitude}").address
#             return latitude, longitude, address
#     except Exception as e:
#         st.error(f"Error retrieving location: {e}")
#     return None, None, None

# # User Login
# st.sidebar.header("User Login")
# user_email = st.sidebar.text_input("Enter your email")
# login_button = st.sidebar.button("Login")

# if login_button:
#     if user_email:
#         st.session_state["user_email"] = user_email
#         st.sidebar.success(f"Logged in as {user_email}")

# if "user_email" not in st.session_state:
#     st.warning("Please log in first.")
#     st.stop()

# # Get user location and display details
# latitude, longitude, address = get_user_location()
# if latitude and longitude:
#     st.success(f"Location detected: {address}")
# else:
#     st.warning("Unable to fetch location, please ensure location access is enabled.")
#     st.stop()

# # Streamlit App
# st.title("BinSight: Upload Dustbin Image")

# uploaded_file = st.file_uploader("Upload an image of the dustbin", type=["jpg", "jpeg", "png"])
# submit_button = st.button("Analyze and Upload")

# if submit_button and uploaded_file:
#     image = Image.open(uploaded_file)
#     st.image(image, caption="Uploaded Image", use_container_width=True)

#     classification_results = classify_image_with_mobilenet(image)
    
#     if classification_results:
#         db_ref = db.reference("dustbins")
#         dustbin_data = {
#             "user_email": st.session_state["user_email"],
#             "latitude": latitude,
#             "longitude": longitude,
#             "address": address,
#             "classification": classification_results,
#             "allocated_truck": None,
#             "status": "Pending"
#         }
#         db_ref.push(dustbin_data)
#         st.success("Dustbin data uploaded successfully!")
#         st.write(f"**Location:** {address}")
#         st.write(f"**Latitude:** {latitude}, **Longitude:** {longitude}")
#     else:
#         st.error("Missing classification details. Cannot upload.")







# best with firebase but below code is not giving correct location of user.

# import streamlit as st
# import requests
# import firebase_admin
# from firebase_admin import credentials, db, auth
# from PIL import Image
# import numpy as np
# from geopy.geocoders import Nominatim
# from tensorflow.keras.applications import MobileNetV2
# from tensorflow.keras.applications.mobilenet_v2 import decode_predictions, preprocess_input

# # Initialize Firebase
# if not firebase_admin._apps:
#     cred = credentials.Certificate("firebase_credentials.json")  
#     firebase_admin.initialize_app(cred, {
#         'databaseURL': 'https://binsight-beda0-default-rtdb.asia-southeast1.firebasedatabase.app/'
#     })

# # Load MobileNetV2 pre-trained model
# mobilenet_model = MobileNetV2(weights="imagenet")

# # Function to classify the uploaded image using MobileNetV2
# def classify_image_with_mobilenet(image):
#     try:
#         img = image.resize((224, 224))
#         img_array = np.array(img)
#         img_array = np.expand_dims(img_array, axis=0)
#         img_array = preprocess_input(img_array)
#         predictions = mobilenet_model.predict(img_array)
#         labels = decode_predictions(predictions, top=5)[0]
#         return {label[1]: float(label[2]) for label in labels}
#     except Exception as e:
#         st.error(f"Error during image classification: {e}")
#         return {}

# # Function to get user's location
# def get_user_location():
#     try:
#         ip_info = requests.get("https://ipinfo.io/json").json()
#         location = ip_info.get("loc", "").split(",")
#         latitude = location[0] if len(location) > 0 else None
#         longitude = location[1] if len(location) > 1 else None

#         if latitude and longitude:
#             geolocator = Nominatim(user_agent="binsight")
#             address = geolocator.reverse(f"{latitude}, {longitude}").address
#             return latitude, longitude, address
#         return None, None, None
#     except Exception as e:
#         st.error(f"Unable to get location: {e}")
#         return None, None, None

# # User Login
# st.sidebar.header("User Login")
# user_email = st.sidebar.text_input("Enter your email")
# login_button = st.sidebar.button("Login")

# if login_button:
#     if user_email:
#         st.session_state["user_email"] = user_email
#         st.sidebar.success(f"Logged in as {user_email}")

# if "user_email" not in st.session_state:
#     st.warning("Please log in first.")
#     st.stop()

# # Streamlit App
# st.title("BinSight: Upload Dustbin Image")

# uploaded_file = st.file_uploader("Upload an image of the dustbin", type=["jpg", "jpeg", "png"])
# submit_button = st.button("Analyze and Upload")

# if submit_button and uploaded_file:
#     image = Image.open(uploaded_file)
#     st.image(image, caption="Uploaded Image", use_container_width=True)

#     classification_results = classify_image_with_mobilenet(image)
#     latitude, longitude, address = get_user_location()

#     if latitude and longitude and classification_results:
#         db_ref = db.reference("dustbins")
#         dustbin_data = {
#             "user_email": st.session_state["user_email"],
#             "latitude": latitude,
#             "longitude": longitude,
#             "address": address,
#             "classification": classification_results,
#             "allocated_truck": None,
#             "status": "Pending"
#         }
#         db_ref.push(dustbin_data)
#         st.success("Dustbin data uploaded successfully!")
#     else:
#         st.error("Missing classification or location details. Cannot upload.")











# Below is the old version but it is without of firebase and here is the addition of gemini.

# import streamlit as st
# import os
# from PIL import Image
# import numpy as np
# from io import BytesIO
# from dotenv import load_dotenv
# from geopy.geocoders import Nominatim
# from tensorflow.keras.applications import MobileNetV2
# from tensorflow.keras.applications.mobilenet_v2 import decode_predictions, preprocess_input
# import requests
# import google.generativeai as genai

# # Load environment variables
# load_dotenv()

# # Configure Generative AI
# genai.configure(api_key='AIzaSyBREh8Uei7uDCbzPaYW2WdalOdjVWcQLAM')

# # Load MobileNetV2 pre-trained model
# mobilenet_model = MobileNetV2(weights="imagenet")

# # Function to classify the uploaded image using MobileNetV2
# def classify_image_with_mobilenet(image):
#     try:
#         img = image.resize((224, 224))
#         img_array = np.array(img)
#         img_array = np.expand_dims(img_array, axis=0)
#         img_array = preprocess_input(img_array)
#         predictions = mobilenet_model.predict(img_array)
#         labels = decode_predictions(predictions, top=5)[0]
#         return {label[1]: float(label[2]) for label in labels}
#     except Exception as e:
#         st.error(f"Error during image classification: {e}")
#         return {}

# # Function to get user's location
# def get_user_location():
#     try:
#         ip_info = requests.get("https://ipinfo.io/json").json()
#         location = ip_info.get("loc", "").split(",")
#         latitude = location[0] if len(location) > 0 else None
#         longitude = location[1] if len(location) > 1 else None

#         if latitude and longitude:
#             geolocator = Nominatim(user_agent="binsight")
#             address = geolocator.reverse(f"{latitude}, {longitude}").address
#             return latitude, longitude, address
#         return None, None, None
#     except Exception as e:
#         st.error(f"Unable to get location: {e}")
#         return None, None, None

# # Function to get nearest municipal details with contact info
# def get_nearest_municipal_details(latitude, longitude):
#     try:
#         if latitude and longitude:
#             # Simulating municipal service retrieval
#             municipal_services = [
#                 {"latitude": "12.9716", "longitude": "77.5946", "office": "Bangalore Municipal Office", "phone": "+91-80-12345678"},
#                 {"latitude": "28.7041", "longitude": "77.1025", "office": "Delhi Municipal Office", "phone": "+91-11-98765432"},
#                 {"latitude": "19.0760", "longitude": "72.8777", "office": "Mumbai Municipal Office", "phone": "+91-22-22334455"},
#             ]

#             # Find the nearest municipal service (mock logic: matching first two decimal points)
#             for service in municipal_services:
#                 if str(latitude).startswith(service["latitude"][:5]) and str(longitude).startswith(service["longitude"][:5]):
#                     return f"""
#                     **Office**: {service['office']}  
#                     **Phone**: {service['phone']}  
#                     """
#             return "No nearby municipal office found. Please check manually."
#         else:
#             return "Location not available. Unable to fetch municipal details."
#     except Exception as e:
#         st.error(f"Unable to fetch municipal details: {e}")
#         return None

# # Function to interact with Generative AI
# def get_genai_response(classification_results, location):
#     try:
#         classification_summary = "\n".join([f"{label}: {score:.2f}" for label, score in classification_results.items()])
#         location_summary = f"""
#         Latitude: {location[0] if location[0] else 'N/A'}
#         Longitude: {location[1] if location[1] else 'N/A'}
#         Address: {location[2] if location[2] else 'N/A'}
#         """
#         prompt = f"""
#         ### You are an environmental expert. Analyze the following:
#         1. **Image Classification**:
#            - {classification_summary}
#         2. **Location**:
#            - {location_summary}

#         ### Output Required:
#         1. Detailed insights about the waste detected in the image.
#         2. Specific health risks associated with the detected waste type.
#         3. Precautions to mitigate these health risks.
#         4. Recommendations for proper disposal.
#         """
#         model = genai.GenerativeModel('gemini-pro')
#         response = model.generate_content(prompt)
#         return response
#     except Exception as e:
#         st.error(f"Error using Generative AI: {e}")
#         return None

# # Function to display Generative AI response
# def display_genai_response(response):
#     st.subheader("Detailed Analysis and Recommendations")
#     if response and response.candidates:
#         response_content = response.candidates[0].content.parts[0].text if response.candidates[0].content.parts else ""
#         st.write(response_content)
#     else:
#         st.write("No response received from Generative AI or quota exceeded.")

# # Streamlit App
# st.title("BinSight: AI-Powered Dustbin and Waste Analysis System")
# st.text("Upload a dustbin image and get AI-powered analysis of the waste and associated health recommendations.")

# uploaded_file = st.file_uploader("Upload an image of the dustbin", type=["jpg", "jpeg", "png"], help="Upload a clear image of a dustbin for analysis.")
# submit_button = st.button("Analyze Dustbin")

# if submit_button:
#     if uploaded_file is not None:
#         image = Image.open(uploaded_file)
#         st.image(image, caption="Uploaded Image", use_container_width =True)

#         # Classify the image using MobileNetV2
#         st.subheader("Image Classification")
#         classification_results = classify_image_with_mobilenet(image)
#         for label, score in classification_results.items():
#             st.write(f"- **{label}**: {score:.2f}")

#         # Get user location
#         location = get_user_location()
#         latitude, longitude, address = location

#         st.subheader("User Location")
#         st.write(f"Latitude: {latitude if latitude else 'N/A'}")
#         st.write(f"Longitude: {longitude if longitude else 'N/A'}")
#         st.write(f"Address: {address if address else 'N/A'}")

#         # Get nearest municipal details with contact info
#         st.subheader("Nearest Municipal Details")
#         municipal_details = get_nearest_municipal_details(latitude, longitude)
#         st.write(municipal_details)

#         # Generate detailed analysis with Generative AI
#         if classification_results:
#             response = get_genai_response(classification_results, location)
#             display_genai_response(response)
#     else:
#         st.write("Please upload an image for analysis.")











# # import streamlit as st
# # import os
# # from PIL import Image
# # import numpy as np
# # from io import BytesIO
# # from dotenv import load_dotenv
# # from geopy.geocoders import Nominatim
# # from tensorflow.keras.applications import MobileNetV2
# # from tensorflow.keras.applications.mobilenet_v2 import decode_predictions, preprocess_input
# # import requests
# # import google.generativeai as genai

# # # Load environment variables
# # load_dotenv()

# # # Configure Generative AI
# # genai.configure(api_key='AIzaSyBREh8Uei7uDCbzPaYW2WdalOdjVWcQLAM')

# # # Load MobileNetV2 pre-trained model
# # mobilenet_model = MobileNetV2(weights="imagenet")

# # # Function to classify the uploaded image using MobileNetV2
# # def classify_image_with_mobilenet(image):
# #     try:
# #         # Resize the image to the input size of MobileNetV2
# #         img = image.resize((224, 224))
# #         img_array = np.array(img)
# #         img_array = np.expand_dims(img_array, axis=0)
# #         img_array = preprocess_input(img_array)

# #         # Predict using the MobileNetV2 model
# #         predictions = mobilenet_model.predict(img_array)
# #         labels = decode_predictions(predictions, top=5)[0]
# #         return {label[1]: float(label[2]) for label in labels}
# #     except Exception as e:
# #         st.error(f"Error during image classification: {e}")
# #         return {}

# # # Function to get user's location
# # def get_user_location():
# #     try:
# #         # Fetch location using the IPInfo API
# #         ip_info = requests.get("https://ipinfo.io/json").json()
# #         location = ip_info.get("loc", "").split(",")
# #         latitude = location[0] if len(location) > 0 else None
# #         longitude = location[1] if len(location) > 1 else None

# #         if latitude and longitude:
# #             geolocator = Nominatim(user_agent="binsight")
# #             address = geolocator.reverse(f"{latitude}, {longitude}").address
# #             return latitude, longitude, address
# #         return None, None, None
# #     except Exception as e:
# #         st.error(f"Unable to get location: {e}")
# #         return None, None, None

# # # Function to get nearest municipal details
# # def get_nearest_municipal_details(latitude, longitude):
# #     try:
# #         if latitude and longitude:
# #             # Simulating municipal service retrieval
# #             return f"The nearest municipal office is at ({latitude}, {longitude}). Please contact your local authority for waste management services."
# #         else:
# #             return "Location not available. Unable to fetch municipal details."
# #     except Exception as e:
# #         st.error(f"Unable to fetch municipal details: {e}")
# #         return None

# # # Function to interact with Generative AI
# # def get_genai_response(classification_results, location):
# #     try:
# #         # Construct prompt for Generative AI
# #         classification_summary = "\n".join([f"{label}: {score:.2f}" for label, score in classification_results.items()])
# #         location_summary = f"""
# #         Latitude: {location[0] if location[0] else 'N/A'}
# #         Longitude: {location[1] if location[1] else 'N/A'}
# #         Address: {location[2] if location[2] else 'N/A'}
# #         """
# #         prompt = f"""
# #         ### You are an environmental expert. Analyze the following:
# #         1. **Image Classification**:
# #            - {classification_summary}
# #         2. **Location**:
# #            - {location_summary}

# #         ### Output Required:
# #         1. Detailed insights about the waste detected in the image.
# #         2. Specific health risks associated with the detected waste type.
# #         3. Precautions to mitigate these health risks.
# #         4. Recommendations for proper disposal.
# #         """

# #         model = genai.GenerativeModel('gemini-pro')
# #         response = model.generate_content(prompt)
# #         return response
# #     except Exception as e:
# #         st.error(f"Error using Generative AI: {e}")
# #         return None

# # # Function to display Generative AI response
# # def display_genai_response(response):
# #     st.subheader("Detailed Analysis and Recommendations")
# #     if response and response.candidates:
# #         response_content = response.candidates[0].content.parts[0].text if response.candidates[0].content.parts else ""
# #         st.write(response_content)
# #     else:
# #         st.write("No response received from Generative AI or quota exceeded.")

# # # Streamlit App
# # st.title("BinSight: AI-Powered Dustbin and Waste Analysis System")
# # st.text("Upload a dustbin image and get AI-powered analysis of the waste and associated health recommendations.")

# # uploaded_file = st.file_uploader("Upload an image of the dustbin", type=["jpg", "jpeg", "png"], help="Upload a clear image of a dustbin for analysis.")
# # submit_button = st.button("Analyze Dustbin")

# # if submit_button:
# #     if uploaded_file is not None:
# #         image = Image.open(uploaded_file)
# #         st.image(image, caption="Uploaded Image", use_column_width=True)

# #         # Classify the image using MobileNetV2
# #         st.subheader("Image Classification")
# #         classification_results = classify_image_with_mobilenet(image)
# #         for label, score in classification_results.items():
# #             st.write(f"- **{label}**: {score:.2f}")

# #         # Get user location
# #         location = get_user_location()
# #         latitude, longitude, address = location

# #         st.subheader("User Location")
# #         st.write(f"Latitude: {latitude if latitude else 'N/A'}")
# #         st.write(f"Longitude: {longitude if longitude else 'N/A'}")
# #         st.write(f"Address: {address if address else 'N/A'}")

# #         # Get nearest municipal details
# #         st.subheader("Nearest Municipal Details")
# #         municipal_details = get_nearest_municipal_details(latitude, longitude)
# #         st.write(municipal_details)

# #         # Generate detailed analysis with Generative AI
# #         if classification_results:
# #             response = get_genai_response(classification_results, location)
# #             display_genai_response(response)
# #     else:
# #         st.write("Please upload an image for analysis.")