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

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

def caption(img):
    inputs = processor(img, return_tensors="pt")
    out = model.generate(**inputs)
    return processor.decode(out[0], skip_special_tokens=True)

demo = gr.Interface(caption, gr.Image(type="pil"), "text")
demo.launch()