app.py

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

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
processor = BlipProcessor.from_pretrained("Salesforce/blip-image-captioning-large")
model = BlipForConditionalGeneration.from_pretrained("Salesforce/blip-image-captioning-large").to(device)

def generate_caption(image):
    # Now directly using the PIL Image object
    inputs = processor(images=image, return_tensors="pt")
    outputs = model.generate(**inputs)
    caption = processor.decode(outputs[0], skip_special_tokens=True, clean_up_tokenization_spaces=True)
    return caption

def caption_image(image):
    """
    Takes a PIL Image input and returns a caption.
    """
    try:
        caption = generate_caption(image)
        return caption
    except Exception as e:
        return f"An error occurred: {str(e)}"

iface = gr.Interface(
    fn=caption_image,
    inputs=gr.Image(type="pil"),
    outputs="text",
    title="Image Captioning with BLIP",
    description="Upload an image to generate a caption."
)

iface.launch(share=True)