app.py

import gradio as gr
import numpy as np
from tensorflow.keras.models import load_model
from tensorflow.keras.preprocessing import image
from PIL import Image

# Load your custom model
model = load_model("lung_cancer_detection_model.h5", compile=False)

# Compile your model using the optimizer
model.compile(optimizer="adam", loss="binary_crossentropy", metrics=["accuracy"])

# Function to preprocess the uploaded image
def predict_single_image(img, model, target_size=(128, 128)):
    # Convert the PIL image to grayscale and resize
    img = img.convert("L")  # Convert to grayscale
    img = img.resize(target_size)  # Resize to target size

    # Convert image to array
    img_array = np.array(img)
    img_array = np.expand_dims(img_array, axis=0)  # Expand dimensions for batch
    img_array = img_array / 255.0  # Normalize the image

    # Predict the class probabilities
    probabilities = model.predict(img_array)

    # Determine the predicted class label
    predicted_class = "positive" if probabilities[0][0] > 0.5 else "negative"

    return predicted_class, probabilities[0][0]

# Function to classify the uploaded image
def classify_lung_cancer(img):
    # Call the function to predict the class label for the single image
    predicted_label, confidence = predict_single_image(img, model, target_size=(128, 128))

    # Return the formatted prediction
    return f"Prediction: {predicted_label}\n(Confidence: {confidence:.2f})"

# Define the Gradio interface
iface = gr.Interface(
    fn=classify_lung_cancer,
    inputs=gr.Image(type="pil"),  # Pass the PIL image directly
    outputs=gr.Textbox(),
    live=True,
    title="Lung Cancer Classification",
    description="Upload an image and the model will classify it as positive or negative for lung cancer.",
)

# Create a Gradio interface
iface.launch(share=True)