app.py

import gradio as gr
import numpy as np
from tensorflow.keras.models import load_model

# Load the trained model
model = load_model('skin_model.h5')

# Define a function to make predictions
def predict(image):
    # Preprocess the image
    image = image / 255.0
    image = np.expand_dims(image, axis=0)
    
    # Make a prediction using the model
    prediction = model.predict(image)
    
    # Get the predicted class label
    if prediction[0][0] < 0.5:
        label = 'Benign'
    else:
        label = 'Malignant'
    
    return label

examples = [["benign.jpg"], ["malignant.jpg"]]

# Customized layout and style for improved UI
interface_layout = [
    gr.Interface(
        fn=predict,
        inputs=gr.inputs.Image(shape=(150, 150)),
        outputs=gr.outputs.Label(),
        examples=examples,
        title="Skin Cancer Classification",
        description="Predicts whether a skin image is cancerous or not.",
        theme="default",  # Choose a theme: "default", "compact", "huggingface"
        layout="vertical",  # Choose a layout: "vertical", "horizontal", "double"
        live=False  # Set to True for live updates without clicking "Submit"
    )
]

gr.Interface(
    layout="colab",  # Choose a layout: "colab", "colab-sandbox", "textbox"
    layout_options={"fullscreen": True},
    interfaces=interface_layout
).launch()