app.py

import streamlit as st
from transformers import pipeline
from PIL import Image
import requests
from io import BytesIO
import logging
import torch

# Setup logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

class AgeClassifier:
    def __init__(self):
        try:
            self.device = "cuda" if torch.cuda.is_available() else "cpu"
            self.pipe = pipeline("image-classification", model="nateraw/vit-age-classifier", device=self.device)
            logger.info(f"Model loaded successfully on {self.device}")
        except Exception as e:
            logger.error(f"Failed to initialize pipeline: {e}")
            raise

    def classify_image(self, image):
        try:
            return self.pipe(image)
        except Exception as e:
            logger.error(f"Classification failed: {e}")
            return None

    @staticmethod
    def format_results(results):
        if not results:
            return "No valid results"
        return results

def load_image_from_url(url):
    try:
        response = requests.get(url, timeout=10)
        response.raise_for_status()
        image = Image.open(BytesIO(response.content))
        return image
    except Exception as e:
        st.error(f"Error loading image from URL: {e}")
        return None

def main():
    st.set_page_config(
        page_title="Age Classification App",
        page_icon="👤",
        layout="wide"
    )

    st.title("Age Classification App 👤")
    st.write("Upload an image or provide a URL to classify the age range of people in the image.")

    # Initialize the classifier
    @st.cache_resource
    def get_classifier():
        return AgeClassifier()

    classifier = get_classifier()

    # Create two columns for input methods
    col1, col2 = st.columns(2)

    with col1:
        st.subheader("Upload Image")
        uploaded_file = st.file_uploader("Choose an image...", type=["jpg", "jpeg", "png"])

    with col2:
        st.subheader("Image URL")
        image_url = st.text_input("Enter image URL")

    # Process the image
    image = None
    if uploaded_file is not None:
        image = Image.open(uploaded_file)
    elif image_url:
        image = load_image_from_url(image_url)

    if image:
        # Display the image
        st.image(image, caption="Input Image", use_column_width=True)

        # Add a classify button
        if st.button("Classify Age"):
            with st.spinner("Classifying..."):
                results = classifier.classify_image(image)
                
                if results:
                    # Create a bar chart
                    st.subheader("Classification Results")
                    
                    # Convert results to format suitable for bar chart
                    labels = [r['label'] for r in results]
                    scores = [r['score'] * 100 for r in results]
                    
                    # Display most likely age range
                    most_likely = max(results, key=lambda x: x['score'])
                    st.success(f"Most likely age range: {most_likely['label']} ({most_likely['score']*100:.1f}%)")
                    
                    # Create bar chart
                    chart_data = {
                        'Age Range': labels,
                        'Confidence (%)': scores
                    }
                    st.bar_chart(chart_data, x='Age Range', y='Confidence (%)')
                    
                    # Display detailed results in an expander
                    with st.expander("See detailed results"):
                        st.write("Confidence scores for all age ranges:")
                        for result in results:
                            st.write(f"{result['label']}: {result['score']*100:.1f}%")
                else:
                    st.error("Could not classify the image. Please try another image.")

    # Add information about the model
    with st.sidebar:
        st.header("About")
        st.write("""
        This app uses the ViT (Vision Transformer) model trained for age classification.
        
        The model classifies images into the following age ranges:
        - 0-2 years
        - 3-9 years
        - 10-19 years
        - 20-29 years
        - 30-39 years
        - 40-49 years
        - 50-59 years
        - 60-69 years
        - 70+ years
        """)
        
        st.write("Model: nateraw/vit-age-classifier")
        st.write(f"Running on: {classifier.device}")

if __name__ == "__main__":
    main()