app.py

# app.py
import streamlit as st
from transformers import BlipProcessor, BlipForConditionalGeneration
from PIL import Image
import requests
from io import BytesIO
from deep_translator import GoogleTranslator
import torch

# Streamlit page config
st.set_page_config(page_title="AI Image Caption Generator", page_icon="🖼️")

st.title("🖼️ AI Image Caption Generator")
st.write("Upload an image or paste a Google Image URL to get multiple captions generated in your preferred language!")

# Language options
languages = {
    "English": "en",
    "Urdu": "ur",
    "Hindi": "hi",
    "French": "fr",
    "Spanish": "es",
    "Arabic": "ar"
}

# Image input options
upload_option = st.radio("Choose Image Input Method:", ("Upload from Computer", "Paste Image URL"))

uploaded_file = None
image_url = None

if upload_option == "Upload from Computer":
    uploaded_file = st.file_uploader("Upload an image", type=["jpg", "png", "jpeg"])
else:
    image_url = st.text_input("Paste Image URL (must be direct link ending with .jpg/.png/.jpeg)")

# Language selection
selected_language = st.selectbox("🌐 Choose Output Language", list(languages.keys()))

# Load image based on input
image = None

if uploaded_file:
    image = Image.open(uploaded_file).convert('RGB')
elif image_url:
    try:
        response = requests.get(image_url)
        if response.status_code == 200:
            image = Image.open(BytesIO(response.content)).convert('RGB')
        else:
            st.error("Failed to fetch image. Please check the URL.")
    except Exception as e:
        st.error(f"Error fetching image: {e}")

if image:
    st.image(image, caption="Selected Image", width=300)

    with st.spinner("Generating captions... please wait ⏳"):
        # Resize for model input
        resized_image = image.resize((384, 384))

        processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-base")
        model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-base")

        inputs = processor(resized_image, return_tensors="pt")

        # Generate multiple captions
        out = model.generate(
            **inputs,
            num_beams=5,
            num_return_sequences=3,
            max_length=50,
            early_stopping=True
        )

        captions = [processor.decode(o, skip_special_tokens=True) for o in out]

    st.success("📝 Captions Generated Successfully!")

    st.subheader(f"Here are the captions in {selected_language}:")

    for idx, cap in enumerate(captions):
        try:
            translated_caption = GoogleTranslator(source='auto', target=languages[selected_language]).translate(cap)
        except Exception as e:
            translated_caption = f"(Translation Error: Showing English) {cap}"

        st.text_area(f"✏️ Caption {idx+1}", value=translated_caption, height=80)

    st.caption("Tip: You can edit, copy, or download captions easily! ✍️")
else:
    st.info("Please upload an image or paste a valid URL to proceed.")