app.py

import gradio as gr

title="Swin Transformer"
description="Gradio Demo for Swin Transformer: Hierarchical Vision Transformer using Shifted Windows. To use it, simply upload your image, or click one of the examples to load them. Read more at the links below."

article = "<p style='text-align: center'><a href='https://arxiv.org/abs/2103.14030' target='_blank'>Swin Transformer: Hierarchical Vision Transformer using Shifted Windows</a> | <a href='https://github.com/microsoft/Swin-Transformer' target='_blank'>Github Repo</a></p>"


io1 = gr.Interface.load("huggingface/microsoft/swin-large-patch4-window12-384-in22k")

io2 = gr.Interface.load("huggingface/microsoft/swin-base-patch4-window7-224-in22k")


def inference(image, model):
    if model == "swin-large-patch4-window12-384-in22k":
        outtext = io1(image)
    else:
        outtext = io2(image)
    return outtext   
    
     
examples=[['tiger.jpeg','swin-large-patch4-window12-384-in22k']]

gr.Interface(
    inference, 
    [gr.inputs.Image(label="Input Image",type='filepath'),gr.inputs.Dropdown(choices=["swin-large-patch4-window12-384-in22k","swin-base-patch4-window7-224-in22k"], type="value", default="swin-large-patch4-window12-384-in22k", label="model")
], 
    gr.outputs.Label(label="Classification"),
    examples=examples,
    article=article,
    title=title,
    description=description).launch(enable_queue=True,cache_examples=True)