app.py

import gradio as gr
from transformers import BlipProcessor, BlipForConditionalGeneration
from PIL import Image
import requests

# Initialize the model and processor
processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-large")
model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-large").to("cuda")

def generate_caption(image):
    # Process the input image
    processed = processor(image, return_tensors="pt").to("cuda")
    # Generate caption
    outputs = model.generate(**processed)
    # Decode and return the first caption
    caption = processor.decode(outputs[0], skip_special_tokens=True)
    return caption

# Gradio interface
iface = gr.Interface(fn=generate_caption, inputs=gr.inputs.Image(type="pil"), outputs="text", title="BLIP Image Captioning")

# Launch the app
iface.launch()