app.py

import io
import os
import streamlit as st
import requests
from PIL import Image
from model import get_caption_model, generate_caption

# Initialize session state for the model
if 'caption_model' not in st.session_state:
    st.session_state.caption_model = get_caption_model()

def predict():
    captions = []
    try:
        pred_caption = generate_caption('tmp.jpg', st.session_state.caption_model)
        st.markdown('#### Predicted Captions:')
        captions.append(pred_caption)

        for _ in range(4):
            pred_caption = generate_caption('tmp.jpg', st.session_state.caption_model, add_noise=True)
            if pred_caption not in captions:
                captions.append(pred_caption)
        
        for c in captions:
            st.write(c)
    except Exception as e:
        st.error(f"An error occurred during prediction: {e}")

st.title('Image Captioner')

# Handle image URL input
img_url = st.text_input(label='Enter Image URL')

if img_url:
    try:
        img = Image.open(requests.get(img_url, stream=True).raw)
        img = img.convert('RGB')
        st.image(img)
        img.save('tmp.jpg')
        predict()
        os.remove('tmp.jpg')
    except Exception as e:
        st.error(f"An error occurred with the image URL: {e}")

st.markdown('<center style="opacity: 70%">OR</center>', unsafe_allow_html=True)

# Handle image file upload
img_upload = st.file_uploader(label='Upload Image', type=['jpg', 'png', 'jpeg'])

if img_upload:
    try:
        img = Image.open(io.BytesIO(img_upload.read()))
        img = img.convert('RGB')
        img.save('tmp.jpg')
        st.image(img)
        predict()
        os.remove('tmp.jpg')
    except Exception as e:
        st.error(f"An error occurred with the image upload: {e}")