app.py

import tensorflow as tf 
import gradio as gr
import numpy as np
from PIL import Image
from tensorflow.keras.models import load_model
# Load your trained model and label map 
path="skinDiseaseDetection (1).h5"
model = load_model(path)  
label_map={0: 'pigmented benign keratosis',
 1: 'melanoma',
 2: 'vascular lesion',
 3: 'actinic keratosis',
 4: 'squamous cell carcinoma',
 5: 'basal cell carcinoma',
 6: 'seborrheic keratosis',
 7: 'dermatofibroma',
 8: 'nevus'}

def predict_image(image):
    """
    Predict the class of an uploaded image of a skin lesion.

    Parameters:
    - image: The uploaded image in PIL format.

    Returns:
    - predicted_class: The name of the predicted class or an error message.
    - confidence: The confidence score of the prediction (0 to 1) or None.
    - prediction_type: The type of prediction or an error message.
    """
    if image is None:
        return "Please upload an image.", None, "No Diagnosis"

    # Preprocess the image
    image = image.resize((100, 75))  # Resize to model's expected input size
    image_array = np.asarray(image)
    image_array = (image_array - np.mean(image_array)) / np.std(image_array)  # Normalize
    image_array = np.expand_dims(image_array, axis=0)  # Add batch dimension

    # Make prediction using the loaded model
    predictions = model.predict(image_array)
    predicted_index = np.argmax(predictions, axis=1)[0]
    confidence = np.max(predictions)
    predicted_class = label_map[predicted_index]

    # Define classes for benign and malignant lesions
    benign_classes = [
        'pigmented benign keratosis', 
        'vascular lesion', 
        'actinic keratosis',
        'seborrheic keratosis', 
        'dermatofibroma', 
        'nevus'
    ]
    malignant_classes = [
        'melanoma', 
        'squamous cell carcinoma', 
        'basal cell carcinoma'
    ]

    # Determine the type of prediction
    if predicted_class in benign_classes:
        prediction_type = 'Benign Neoplasm'
    elif predicted_class in malignant_classes:
        prediction_type = 'Malignant Neoplasm'
    else:
        prediction_type = 'Unknown Neoplasm'

    return predicted_class, float(f"{confidence:.2f}"), prediction_type  # Format confidence here

# Example images and their descriptions
examples = [
    ["bcc.jpg"],
    ["d.jpg"],
    ["m.PNG"],
    ["sck.jpg"]
]

# Set up the Gradio interface
iface = gr.Interface(
    fn=predict_image,
    inputs=gr.Image(type="pil", label="Upload Skin Lesion Image"),
    outputs=[
        gr.Label(label="Predicted Class"),
        gr.Number(label="Confidence Score", precision=2),
        gr.Textbox(label="Diagnosis Type", interactive=False)
    ],
    title="Skin Cancer Image Classification",
    description="Upload an image of a skin lesion to predict its type and determine if the diagnosis is Benign or Malignant.",
    theme="default",  # Use a valid theme like "default" or "compact"
    allow_flagging="never",  # Optional: Disable flagging for a cleaner UI
    examples=examples ,
    css="""
    .gradio-container {
        font-family: 'Arial', sans-serif;
        background-color: #ddebf7;
        border-radius: 10px;
        padding: 20px;
    }
    .gr-button {
        background-color: #4CAF50;
        color: white;
    }
    .gr-image {
        border: 2px solid #ddd;
        border-radius: 10px;
        box-shadow: 0 4px 8px rgba(0, 0, 0, 0.2);
    }
    .output-textbox {
        font-size: 1.2em;
        text-align: center;
        color: #333;
    }
    """,
)

# Launch the Gradio interface
iface.launch()