# Step 1: Install Required Libraries
# pip install streamlit tensorflow pillow huggingface_hub

# Step 2: Import Necessary Libraries
import streamlit as st
import tensorflow as tf
from tensorflow.keras.preprocessing.image import load_img, img_to_array
from huggingface_hub import hf_hub_download
import numpy as np
import os

# Step 3: Download Model from Hugging Face
@st.cache_resource
def load_model():
    model_path = hf_hub_download(repo_id="krishnamishra8848/Brain_Tumor_Detection_Model", filename="brain_tumor_model.h5")
    return tf.keras.models.load_model(model_path)

model = load_model()

# Step 4: Define Prediction Function
def predict_image(image_path):
    IMG_HEIGHT, IMG_WIDTH = 128, 128  # Ensure this matches your model's input size
    img = load_img(image_path, target_size=(IMG_HEIGHT, IMG_WIDTH))  # Resize image
    img_array = img_to_array(img)  # Convert image to array
    img_array = np.expand_dims(img_array, axis=0)  # Add batch dimension
    img_array = img_array / 255.0  # Normalize pixel values to [0, 1]
    prediction = model.predict(img_array)[0][0]
    confidence = prediction if prediction > 0.5 else 1 - prediction
    return prediction, confidence

# Step 5: Streamlit Interface
st.title("Brain Tumor Detection")
st.write("Upload a CT scan image to classify it as Healthy or Tumor with confidence score.")

uploaded_file = st.file_uploader("Upload CT Scan Image", type=["png", "jpg", "jpeg"])

if uploaded_file is not None:
    # Save the uploaded file to a temporary location
    temp_file_path = os.path.join("temp_image.jpg")
    with open(temp_file_path, "wb") as f:
        f.write(uploaded_file.getbuffer())
    
    # Make prediction
    prediction, confidence = predict_image(temp_file_path)
    os.remove(temp_file_path)  # Clean up the temporary file
    
    # Convert confidence to percentage
    confidence_percentage = confidence * 100

    # Display the result
    if prediction > 0.5:
        st.markdown(
            f"<h2 style='color:red;'>Prediction: Tumor Detected with {confidence_percentage:.2f}% confidence.</h2>", 
            unsafe_allow_html=True
        )
        st.write("**Note:** Consult a medical professional for further evaluation.")
    else:
        st.markdown(
            f"<h2 style='color:green;'>Prediction: Healthy with {confidence_percentage:.2f}% confidence.</h2>", 
            unsafe_allow_html=True
        )
        st.write("**Note:** The result suggests a healthy scan. Regular checkups are recommended.")