app.py

import gradio as gr
from transformers import pipeline
import numpy as np
from PIL import Image
import tensorflow as tf

model_path = r'./model1_acc96_kera.h5'

# Function to preprocess the image and make predictions
def classify_alzheimers_image(input_image):
    try:
        # Preprocess the image (resize, normalize, etc.)
        input_image = np.array(input_image)
        input_image_copy = input_image.copy()  # Making a copy to avoid the array reference issue
        input_image_resized = np.array(Image.fromarray(input_image_copy).resize((128, 128))) / 255.0
        input_image_resized = np.expand_dims(input_image_resized, axis=0)

        # Making predictions
        model_h5 = tf.keras.models.load_model(model_path)
        predictions = model_h5.predict(input_image_resized)

        # Getting the class with the highest probability
        class_idx = np.argmax(predictions)
        class_label = ["Mild Demented", "Moderate Demented", "Non-Demented", "Very Mild Demented"][class_idx]
        confidence = predictions[0][class_idx]

        return f"Prediction: {class_label}, Confidence: {confidence:.2f}"
    except Exception as e:
        return str(e)

# Creating a Gradio interface
iface = gr.Interface(
    fn=classify_alzheimers_image,
    inputs="image",
    outputs="text",
    title="Alzheimer's Disease Classification",
    description=" Upload an MRI Image for classification.",
    flagging_options = ["Wrong Prediction"],
    theme = 'darkhuggingface'
)

# Launching the Gradio interface
iface.launch(inline = False)