Yugesh-S/Image-Description-Generator-App

1import streamlit as st2from transformers import BlipProcessor, BlipForConditionalGeneration3from PIL import Image4import torch5 6# Initialize the BLIP model and processor7processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-base")8model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base")9 10# Function to interact with the model and get the description11def get_image_description(image, prompt):12    inputs = processor(image, prompt, return_tensors="pt")13    out = model.generate(**inputs)14    description = processor.decode(out[0], skip_special_tokens=True)15    return description16 17# Streamlit UI18st.title("Image Description using Hugging Face Models")19 20# File uploader for image21uploaded_image = st.file_uploader("Upload an image", type=["jpg", "jpeg", "png"])22 23# Text input for prompt24prompt = st.text_input("Enter your prompt", value="Describe this image")25 26if uploaded_image is not None:27    # Open and resize the uploaded image28    image = Image.open(uploaded_image)29    resized_image = image.resize((400, 400))  # Resize the image to 400x400 pixels30    31    # Display image and description side by side using columns32    col1, col2 = st.columns([2, 3])33    34    with col1:35        # Display the resized image on the left36        st.image(resized_image, caption="Uploaded Image", use_column_width=True)37 38    with col2:39        # Display the description on the right40        if st.button("Get Description"):41            description = get_image_description(image, prompt)42            st.info(description)  # Display answer in an info box43