app.py

import streamlit as st
import numpy as np
import tensorflow as tf
from tensorflow.keras.models import load_model
from PIL import Image
import requests
import tempfile
import os

def download_model_from_gdrive():
    """Downloads the model from Google Drive."""
    gdrive_url = "https://drive.google.com/uc?id=1EnokggrC6ymrSibtj2t7IHWb9QVtrehS"
    output_path = "final_meta_model.keras"

    with requests.get(gdrive_url, stream=True) as r:
        r.raise_for_status()
        with open(output_path, "wb") as f:
            for chunk in r.iter_content(chunk_size=8192):
                f.write(chunk)

    return output_path

def load_or_download_model():
    """Loads the model from Hugging Face or Google Drive."""
    hf_model_path = "final_meta_model.keras"

    # Check if model exists locally (from Hugging Face Space storage)
    if os.path.exists(hf_model_path):
        model = load_model(hf_model_path)
    else:
        st.write("Model not found locally. Downloading from Google Drive...")
        model_path = download_model_from_gdrive()
        model = load_model(model_path)

    return model

def preprocess_image(image):
    """Preprocesses the uploaded image for model prediction."""
    image = image.resize((224, 224))  # Assuming input size of 224x224 for the model
    image_array = np.array(image)
    image_array = image_array / 255.0  # Normalize pixel values to [0, 1]
    image_array = np.expand_dims(image_array, axis=0)  # Add batch dimension
    return image_array

def predict_severity(model, image):
    """Predicts the severity using the model."""
    predictions = model.predict(image)
    class_names = ['0', 'MR', 'MRMS', 'MS', 'R', 'S']
    predicted_class = class_names[np.argmax(predictions)]
    return predicted_class, predictions

# Streamlit App
st.title("Disease Severity Prediction App")
st.write("Upload an image to predict the severity of the disease.")

# Image Upload
uploaded_image = st.file_uploader("Upload an image", type=["jpg", "jpeg", "png"])

if uploaded_image:
    st.image(uploaded_image, caption="Uploaded Image", use_column_width=True)

    # Load model
    with st.spinner("Loading model..."):
        model = load_or_download_model()

    # Preprocess image
    image = Image.open(uploaded_image)
    preprocessed_image = preprocess_image(image)

    # Predict severity
    with st.spinner("Predicting severity..."):
        predicted_class, prediction_scores = predict_severity(model, preprocessed_image)

    # Display results
    st.success(f"Predicted Class: {predicted_class}")
    st.write("Prediction Scores:")
    for class_name, score in zip(['0', 'MR', 'MRMS', 'MS', 'R', 'S'], prediction_scores[0]):
        st.write(f"{class_name}: {score:.4f}")