app.py

import cv2
import numpy as np
import tensorflow as tf
import gradio as gr

# Load the trained model
model = tf.keras.models.load_model('unet_model.h5')

# Function to preprocess the images
def preprocess_images(old_image, current_image, img_size=128):
    old_image_resized = cv2.resize(old_image, (img_size, img_size)) / 255.0
    current_image_resized = cv2.resize(current_image, (img_size, img_size)) / 255.0
    input_combined = np.concatenate([old_image_resized, current_image_resized], axis=-1)
    input_combined = np.expand_dims(input_combined, axis=0)
    return input_combined

# Prediction function
def predict_mask(old_image, current_image):
    # Convert from PIL to numpy array
    old_image_np = np.array(old_image)
    current_image_np = np.array(current_image)

    # Preprocess images
    preprocessed_input = preprocess_images(old_image_np, current_image_np)

    # Predict the mask
    prediction = model.predict(preprocessed_input)
    prediction_mask = (prediction[0] > 0.5).astype(np.uint8) * 255  # Binary mask

    # Resize mask back to original size
    mask_resized = cv2.resize(prediction_mask, (old_image_np.shape[1], old_image_np.shape[0]))

    return mask_resized

# Gradio Interface
interface = gr.Interface(
    fn=predict_mask,
    inputs=[
        gr.Image(label="Old Image"),
        gr.Image(label="Current Image")
    ],
    outputs=gr.Image(label="Predicted Mask"),
    title="Change Detection with U-Net",
    description="Upload two images (old and current) to detect changes using a U-Net model."
)

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