app.py

import gradio as gr
from utils import get_captions

# Create a Gradio interface
iface = gr.Interface(
    fn=get_captions,  # Function to be called on user input
    inputs=gr.Image(
            width=416, height=416, 
            type="pil", image_mode='RGB', label="Upload Image"
        ),
    outputs=gr.Textbox(
            label="Response from AI Model: ", 
        ),
    examples = ['examples/HorseRide.jpg','examples/Car.jpg']
)

# Launch the Gradio app
iface.launch()