from transformers import pipeline
import streamlit as st

pipe = pipeline("image-to-text", model="nlpconnect/vit-gpt2-image-captioning")

url = st.text_area("enter image address:")

if url:
    out = pipe(url)
    st.json(out)