app.py

import gradio as gr
from gradio_client import Client

client = Client("vikhyatk/moondream1")

result = client.predict(
    "https://hips.hearstapps.com/hmg-prod/images/dog-puppy-on-garden-royalty-free-image-1586966191.jpg",
    "What is this image? Describe this image to someone who is visually impaired.",
    api_name="/answer_question",
)

def get_caption(image, additional_context):
    question_with_context = "What is this image? Describe this image to someone who is visually impaired. " + additional_context
    result = client.predict(image, question_with_context, api_name="/answer_question")
    return result

with gr.Blocks() as demo:
    with gr.Row():
        image = gr.Image(sources=["upload", "clipboard"], type="filepath", height=300)
        additional_context = gr.Textbox(interactive=True, label="Add additional context here (optional)")
    with gr.Row():
        response = gr.Textbox(label="Answer")
    submit_btn = gr.Button("Submit", variant="primary")

    submit_btn.click(get_caption, [image, additional_context], outputs=response)
    
if __name__ == "__main__":
    demo.launch()