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'
        confidence = 1 - prediction[0][0]  # Confidence for benign
    else:
        label = 'Malignant'
        confidence = prediction[0][0]  # Confidence for malignant
    
    return {'label': label, 'confidence': float(confidence)}  # Convert confidence to float

# Custom post-processing function to sort by confidence
def custom_postprocess(output):
    sorted_output = sorted(output.items(), key=lambda x: x[1], reverse=True)
    return f"{sorted_output[0][0]} ({sorted_output[0][1] * 100:.2f}%)"

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

# Define input and output components
image_input = gr.inputs.Image(shape=(150, 150))
label_output = gr.outputs.Label(postprocess=custom_postprocess)

# Define a Gradio interface for user interaction
iface = gr.Interface(
    fn=predict,
    inputs=image_input,
    outputs=label_output,
    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"
)

iface.launch()