app.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

import torch
import gradio as gr
from lib import create_model
from lib.options import ParamSet, _retrieve_parameter, _dispatch_by_group
from lib.dataloader import ImageMixin


test_weight = './weight_epoch-200_best.pt'
parameter = './parameters.json'

class ImageHandler(ImageMixin):
    def __init__(self, params):
        self.params = params
        self.transform = self._make_transforms()

    def set_image(self, image):
        image = self.transform(image)
        image = {'image': image.unsqueeze(0)}
        return image

def load_parameter(parameter):
    _args = ParamSet()
    params = _retrieve_parameter(parameter)
    for _param, _arg in params.items():
        setattr(_args, _param, _arg)

    _args.augmentation = 'no'
    _args.sampler = 'no'
    _args.pretrained = False
    _args.mlp = None
    _args.net = _args.model
    _args.device = torch.device('cpu')

    args_model = _dispatch_by_group(_args, 'model')
    args_dataloader = _dispatch_by_group(_args, 'dataloader')
    return args_model, args_dataloader

args_model, args_dataloader = load_parameter(parameter)
model = create_model(args_model)
model.load_weight(test_weight)

def main(image):
    model.eval()
    image_handler = ImageHandler(args_dataloader)
    image = image_handler.set_image(image)

    with torch.no_grad():
        outputs = model(image)

    label_name = list(outputs.keys())[0]
    result = outputs[label_name].detach().numpy().item()
    result = f"{result:.2f}"
    return result

html_content = """
<div style="padding: 15px; border: 1px solid #e0e0e0; border-radius: 5px;">
    <h3>Image preprocess</h3>
    <p>Only grayscale 320×320 resolution works appropriately.</p>
    <p>The longest side of the Xp was downscaled to 320 pixels while maintaining the aspect ratio, 
        and the width along the shorter side was then padded black to 320 pixels.
    </p>
    <h3>Publication Details</h3>
    <p>See details in our publication, titled 
        "Chest radiography as a biomarker of ageing: artificial intelligence-based, 
        multi-institutional model development and validation in Japan"
    </p>
    <p><strong>Link:</strong> <a href="https://www.thelancet.com/journals/lanhl/article/PIIS2666-7568(23)00133-2/fulltext" target="_blank">
        https://www.thelancet.com/journals/lanhl/article/PIIS2666-7568(23)00133-2/fulltext
    </a></p>
</div>
"""
# Gradio
with gr.Blocks(title="Aging Biomarker from CXR",
            css=".gradio-container {background:mintcream;}"
        ) as demo:
    gr.HTML("""<div style="text-align:center"><h2>Aging Biomarker from CXR</h2></div>""")
    gr.HTML(html_content)
            
    with gr.Row(): 
        input_image = gr.Image(type="pil", image_mode="L", shape=(320, 320))        
        output_label=gr.Label(label="Estimated age")

    send_btn = gr.Button("Inference")
    send_btn.click(fn=main, inputs=input_image, outputs=output_label)

    with gr.Row():
        gr.Examples(['./samples/66_female_xp.png'], label='Sample CXR 1: 66 years old female', inputs=input_image)
        gr.Examples(['./samples/28_male_xp.png'], label='Sample CXR 2: 28 years old male', inputs=input_image)

demo.launch(debug=True)