import streamlit as st
import PIL.Image as Image
import numpy as np
import pandas as pd
import requests
from io import BytesIO
from fastai.vision.all import load_learner  

# Initialize Streamlit app
st.title("White Blood Cell Classifier")

# Add a description or subtitle
st.markdown("""
This app allows you to classify white blood cells from an uploaded image. 
You can upload an image of a blood sample, and the app will predict the type of white blood cell present. 
Choose from various cell types like eosinophil, lymphocyte, monocyte, and neutrophil.
Note: To get the best results, please make sure there is only one WBC in the image. This model has not been trained on basophils.
""")

# Load the FastAI model for WBC identification
fastai_model = load_learner('model1.pkl')

# File uploader for image input
uploaded_file = st.file_uploader("Upload an image for classification", type=["jpg", "png"])

if uploaded_file:
    # Open the uploaded image
    image = Image.open(uploaded_file).convert('RGB')

    
    # Display the uploaded image with a caption
    st.image(image, caption="Reduced Size Image", use_column_width=False, width=150)  # 150 pixels wide

    # Perform inference with the FastAI model
    pred, idx, probs = fastai_model.predict(image)

    # Display a title for the results section
    st.subheader("White Blood Cell Classification Results")

    # Define categories for classification
    categories = ('EOSINOPHIL', 'LYMPHOCYTE', 'MONOCYTE', 'NEUTROPHIL')

    # Create a DataFrame with classification probabilities
    results_df = pd.DataFrame(
        {'Cell Type': categories, 'Probability': probs.tolist()}
    )
     # Highlight the most likely class
    most_likely_class = categories[idx]
    st.success(f"Predicted Class: {most_likely_class}")

    # Additional information about the probabilities
    st.write("Detailed Classification Results:")
    st.table(results_df)

    # Display the probabilities as a bar chart
    st.bar_chart(results_df.set_index('Cell Type'))

   
else:
    st.warning("Upload an image to start classification.")