upload.py

import streamlit as st
import tensorflow as tf
import numpy as np
import cv2
from PIL import Image
import matplotlib.pyplot as plt
from datetime import datetime

# Load your model
model = tf.keras.models.load_model("SkinCancer.h5")
class_names = [
    "Actinic keratoses and intraepithelial carcinomae",
    "Basal Cell Carcinoma",
    "Benign Keratosis-like Lesions",
    "Dermatofibroma",
    "Melanocytic Nevi",
    "Pyogenic Granulomas and Hemorrhage",
    "Melanoma"
]

def create_report_image(name, phone, age, email, prediction, confidence, uploaded_image):
    plt.figure(figsize=(8, 10))

    # Create report text in a structured format
    report_text = (
        f"Skin Disease Prediction Report\n"
        f"Date: {datetime.now().strftime('%Y-%m-%d')}\n\n"
        f"Patient Details:\n"
        f"Name: {name}\n"
        f"Phone Number: {phone}\n"
        f"Age: {age}\n"
        f"Email: {email}\n\n"
        f"Predicted Disease: {prediction}\n"
        f"Confidence: {confidence:.2f}%\n\n"
        f"This report is generated by an AI model. Please consult a healthcare professional for a further diagnosis."
    )

    # Add report text to the top
    plt.subplot(2, 1, 1)  # Two rows, one column, first subplot
    plt.axis('off')  # Hide axes
    plt.text(0.5, 0.5, report_text, fontsize=12, ha='center', va='center', wrap=True)
    plt.title("Prediction Report", fontsize=16)

    # Display the uploaded image at the bottom with a heading
    plt.subplot(2, 1, 2)  # Two rows, one column, second subplot
    plt.axis('off')
    plt.title("User's Uploaded Image", fontsize=14)
    plt.imshow(uploaded_image)

    plt.tight_layout()  # Adjust layout to prevent overlap
    plt.savefig("prediction_report.png", bbox_inches='tight')
    plt.close()

def app():
    st.title("Skin Disease Prediction⏳")
    
    # Display the header image
    st.image("https://cdn.medicalfuturist.com/wp-content/uploads/2023/09/tmf_article_386.png", use_column_width=True)
    
    # Image upload
    uploaded_file = st.file_uploader("**Upload an image of the skin lesion** 👇🏻", type=["jpg", "jpeg", "png"])
    
    if uploaded_file is not None:
        image = Image.open(uploaded_file)
        st.image(image, caption='Uploaded Image', use_column_width=True)

        # User details input
        st.subheader("Please enter your details:👇")
        name = st.text_input("Name")
        phone = st.text_input("Phone Number", max_chars=10)
        age = st.number_input("Age", min_value=0, max_value=120)
        email = st.text_input("Email")

        if st.button("Predict"):
            # Validation
            if len(name) < 5:
                st.error("Name must be at least 5 characters long.")
                return
            
            if not (phone.isdigit() and len(phone) >= 10):
                st.error("Phone number must be exactly 10 digits.")
                return

            if "gmail.com" not in email:
                st.error("Email must be a Gmail address.")
                return

            def predict_skin_disease(image, model):
                image = np.array(image)
                resized_image = cv2.resize(image, (28, 28))
                image_batch = np.expand_dims(resized_image, axis=0)

                pred = model.predict(image_batch)
                predicted_class = np.argmax(pred, axis=1)
                confidence = np.max(pred) * 100  # Confidence percentage
                return class_names[predicted_class[0]], confidence

            prediction, confidence = predict_skin_disease(image, model)
            st.success(f"Prediction: {prediction} (Confidence: {confidence:.2f}%)")
            
            # Create columns for user details and report
            col1, col2 = st.columns([1, 2])  # Adjust column widths as needed
            
            with col1:
                # Display user details
                st.subheader("User Details:")
                st.write(f"Name: {name}")
                st.write(f"Phone Number: {phone}")
                st.write(f"Age: {age}")
                st.write(f"Email: {email}")

            with col2:
                # Provide report
                st.subheader("Prediction Report📑")
                st.markdown(f"The model predicts that the skin lesion is most likely: **{prediction}** with a confidence of **{confidence:.2f}%**.")
                st.write("Please consult a healthcare professional for a further diagnosis.")

                # Generate report image
                create_report_image(name, phone, age, email, prediction, confidence, image)

                # Center display of the report image
                st.markdown("<h3 style='text-align: center;'>Prediction Report</h3>", unsafe_allow_html=True)
                st.image("prediction_report.png", use_column_width=True)

if __name__ == "__main__":
    app()