app.py

import gradio as gr
import tensorflow as tf
import numpy as np
from PIL import Image
import os

# Load the trained model
model = tf.keras.models.load_model("eff_model.h5")

# Same normalization you used in training
def preprocess_image(image: Image.Image):
    image = image.resize((512, 512)).convert("RGB")
    image = np.array(image).astype(np.float32) / 255.0

    mean = np.array([0.44101639, 0.45513914, 0.40195001])
    std = np.array([0.28792392, 0.29775171, 0.29840153])

    image = (image - mean) / std
    image = np.expand_dims(image, axis=0)  # Add batch dimension
    return image

def predict(image: Image.Image):
    processed = preprocess_image(image)
    prediction = model.predict(processed)[0][0]  # sigmoid output

    label = "🌋 Volcanic Eruption" if prediction > 0.5 else "✅ No Eruption"
    confidence = f"{prediction:.2%}" if prediction > 0.5 else f"{(1 - prediction):.2%}"
    return f"{label} (Confidence: {confidence})"

# Gradio Interface
interface = gr.Interface(
    fn=predict,
    inputs=gr.Image(type="pil", label="Upload Satellite Image"),
    outputs=gr.Textbox(label="Prediction"),
    title="Volcanic Eruption Detection",
    description="Upload a satellite image to detect a volcanic eruption using EfficientNetB7."
)

if __name__ == "__main__":
    interface.launch()