import gradio
from transformers import pipeline


_pipeline = pipeline("image-to-text", "Salesforce/blip-image-captioning-base")


def launch(pipeline_input):
    out = _pipeline(pipeline_input)
    return out[0]["generated_text"]


_interface = gradio.Interface(launch, inputs=gradio.Image(type="pil"), outputs="text")

_interface.launch()