app.py

import streamlit as st
import tensorflow as tf
from io import BytesIO
import numpy as np
import cv2
import base64

def set_page_background(png_file):
    @st.cache_data(show_spinner=False)
    def get_base64_of_bin_file(bin_file):
        with open(bin_file, 'rb') as f:
            data = f.read()
        return base64.b64encode(data).decode()

    bin_str = get_base64_of_bin_file(png_file)
    custom_css = f'''
        <style>
            .stApp {{
                background-image: url("data:image/png;base64,{bin_str}");
                background-size: cover;
                background-repeat: no-repeat;
                background-attachment: scroll;
            }}
            
            #MainMenu {{visibility: hidden;}}
            footer {{visibility: hidden;}}
            icon {{color: white;}}
            nav-link {{--hover-color: grey; }}
            nav-link-selected {{background-color: #4ABF7E;}}
        </style>
    '''
    st.markdown(custom_css, unsafe_allow_html=True)

set_page_background("./technology-network-background-connection-cyber-space-ai-generative.jpg")
st.title("Stages of Alzheimer's Disease Prediction")

st.markdown("[Dataset Source](https://www.kaggle.com/datasets/tourist55/alzheimers-dataset-4-class-of-images)")

model = tf.keras.models.load_model('./model/model_1.h5')
model.load_weights('./model/best_model_custom_1.h5')

uploaded_file = st.file_uploader("Upload a brain MRI image here", type=["jpg", "png", "jpeg"], label_visibility="hidden")
predict_button = st.button("ㅤㅤPredictㅤㅤ")

if uploaded_file is not None:
    file_bytes = BytesIO(uploaded_file.read())
    st.image(file_bytes,use_column_width=True,clamp = True) 
    
    img = cv2.imdecode(np.frombuffer(file_bytes.read(), np.uint8), 0)
    #img=np.array(file_bytes)   
  
    if len(img.shape) == 2:
        img=cv2.cvtColor(img,cv2.COLOR_GRAY2RGB)
    
    img=cv2.resize(img,(176, 176))
    if img.max() > 1:
        img = img / 255.0
    
    img = np.expand_dims(img, axis=0)
    pred=model.predict(img)
    predict_val = np.argmax(pred, axis=1)
   
    if predict_val == 0:
        probability = pred[predict_val]
        st.write(f"Mildly Demented with prediction probability of {probability}") 
                 
    elif predict_val == 1:
        probability = pred[predict_val]
        st.write(f"Moderately Demented with prediction probability of {probability}") 
                 
    elif predict_val == 2:
        probability = pred[predict_val]
        st.write(f"Not Demented with prediction probability of {probability}") 
                
    elif predict_val == 3:
        probability = pred[predict_val]
        st.write(f"Very Mildly Demented with prediction probability of {probability}") 
                
    else:
       st.write("Error!")