app.py

import numpy as np
from PIL import Image
import streamlit as st
from transformers import GPT2Tokenizer, GPT2LMHeadModel
from transformers import AutoTokenizer, VisionEncoderDecoderModel, ViTFeatureExtractor

# Load the Model,feature extractor and tokenizer
model = VisionEncoderDecoderModel.from_pretrained("nlpconnect/vit-gpt2-image-captioning") 
extractor = ViTFeatureExtractor.from_pretrained("nlpconnect/vit-gpt2-image-captioning")
tokeniser = AutoTokenizer.from_pretrained("nlpconnect/vit-gpt2-image-captioning")

def generate_captions(image):
    generated_caption = tokeniser.decode(model.generate(extractor(image, return_tensors="pt").pixel_values.to("cpu"))[0])
    sentence = generated_caption
    text_to_remove = "<|endoftext|>"
    generated_caption = sentence.replace(text_to_remove, "")
    return generated_caption

# Load the pre-trained model and tokenizer
model_name = "gpt2"
tokenizer = GPT2Tokenizer.from_pretrained(model_name)
model = GPT2LMHeadModel.from_pretrained(model_name)

# Define the Streamlit app
def generate_paragraph(prompt):
    # Tokenize the prompt
    input_ids = tokenizer.encode(prompt, return_tensors="pt")

    # Generate the paragraph
    output = model.generate(input_ids, max_length=200, num_return_sequences=1, early_stopping=True)

    # Decode the generated output into text
    paragraph = tokenizer.decode(output[0], skip_special_tokens=True)
    return paragraph

# Streamlit app
def main():
    # Set Streamlit app title and description
    st.title("Paragraph Generation From Context of an Image")
    st.subheader("Upload the Image to generate a paragraph.")

    # create file uploader
    uploaded_file  = st.file_uploader("Choose an image...", type=["jpg", "jpeg", "png"])

    # check if file has been uploaded
    if uploaded_file is not None:
        # load the image
        image = Image.open(uploaded_file).convert("RGB")
        
        # context as prompt
        prompt = generate_captions(image)
        st.write("The Context is:", prompt)

        # display the image
        st.image(uploaded_file)
    
        generated_paragraph = generate_paragraph(prompt)

        st.write(generated_paragraph)

if __name__ == "__main__":
    main()