app.py

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

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model = BlipForConditionalGeneration.from_pretrained("./").to(device)
processor = BlipProcessor.from_pretrained("./")

def generate_caption(image):
    inputs = processor(images=image, return_tensors="pt").to(device)
    with torch.no_grad():
        generated_ids = model.generate(**inputs)
    caption = processor.decode(generated_ids[0], skip_special_tokens=True)
    return caption

interface = gr.Interface(fn=generate_caption, inputs="image", outputs="text")
interface.launch()