predict.py

import albumentations as A
import gradio as gr
import numpy as np
import torch
from albumentations.pytorch import ToTensorV2
from models.model_zoo import BoneAgeEstModelZoo
device = "cpu"


def initialize_model():
    # Load model
    model = BoneAgeEstModelZoo(branch="gender", pretrained=True, lr=0.001).load_from_checkpoint(
        "/bae/output/inception_1024_new_data/epoch=13-step=9828.ckpt", map_location=device)
    model.model.eval()
    model.classifier.eval()
    model.gender.eval()
    return model


# Preprocessing and postprocessing
transform = A.Compose([
    A.Resize(width=1024, height=1024),
    A.CLAHE(),
    A.Normalize(),
    ToTensorV2(),
])


def predict(image, gender):
    if gender:
        is_female = 1
    else:
        is_female = 0

    model = initialize_model()

    processed_image = transform(image=np.array(image, dtype=np.uint8))['image']
    processed_image = processed_image.unsqueeze(0)
    processed_image = processed_image.to(device)
    is_female = torch.tensor(is_female).unsqueeze(0).unsqueeze(1).to(device)

    scans = {
        'image': processed_image,
        'gender': is_female
    }
    preds = model(scans)
    return int(preds)


def run():
    image_input = gr.inputs.Image(type="pil", label="Input PNG image")
    gender_input = gr.Checkbox(label="Is Female?", info="Is the scan of a female?", default=True)
    output = gr.outputs.Textbox(label="Predicted Age")

    BAE = gr.Interface(
        fn=predict,
        inputs=[image_input, gender_input],
        outputs=output,
    )

    BAE.launch(server_name="0.0.0.0", server_port=7860, debug=True, show_error=True)


if __name__ == "__main__":
    run()