app.py

from pathlib import Path

import gradio as gr
import numpy as np
import onnxruntime as ort
from huggingface_hub import hf_hub_download
from PIL import Image


MODEL_REPO = "ReaganWZY/DepthPolyp"
ONNX_FILENAME = "DepthPolyp_Kvasir.onnx"
IMAGE_SIZE = 224
EXAMPLES_DIR = Path("samples/kvasir/images")

SESSION = None
INPUT_NAME = None


def get_session():
    global SESSION, INPUT_NAME
    if SESSION is None:
        model_path = hf_hub_download(repo_id=MODEL_REPO, filename=ONNX_FILENAME)
        SESSION = ort.InferenceSession(model_path, providers=["CPUExecutionProvider"])
        INPUT_NAME = SESSION.get_inputs()[0].name
    return SESSION, INPUT_NAME


def preprocess(image: Image.Image):
    image = image.convert("RGB")
    original_size = image.size
    resized = image.resize((IMAGE_SIZE, IMAGE_SIZE), Image.BILINEAR)
    array = np.asarray(resized).astype(np.float32) / 255.0
    tensor = np.transpose(array, (2, 0, 1))[None, ...]
    return image, original_size, tensor


def to_grayscale(probability: np.ndarray, size):
    probability = np.clip(probability, 0.0, 1.0)
    image = Image.fromarray((probability * 255).astype(np.uint8), mode="L")
    return image.resize(size, Image.BILINEAR)


def colorize_purple_yellow(probability: np.ndarray, size):
    probability = np.clip(probability, 0.0, 1.0)
    stops = np.array(
        [
            [38, 5, 84],
            [86, 33, 132],
            [141, 48, 140],
            [203, 71, 119],
            [245, 135, 48],
            [252, 231, 37],
        ],
        dtype=np.float32,
    )
    scaled = probability * (len(stops) - 1)
    lower = np.floor(scaled).astype(np.int32)
    upper = np.clip(lower + 1, 0, len(stops) - 1)
    alpha = (scaled - lower)[..., None]
    colored = stops[lower] * (1.0 - alpha) + stops[upper] * alpha
    image = Image.fromarray(colored.astype(np.uint8), mode="RGB")
    return image.resize(size, Image.BILINEAR)


def make_overlay(image: Image.Image, mask_probability: Image.Image):
    base = image.convert("RGBA")
    mask_array = np.asarray(mask_probability).astype(np.float32) / 255.0
    color = np.zeros((mask_array.shape[0], mask_array.shape[1], 4), dtype=np.uint8)
    color[..., 0] = 252
    color[..., 1] = 231
    color[..., 2] = 37
    color[..., 3] = (mask_array * 155).astype(np.uint8)
    return Image.alpha_composite(base, Image.fromarray(color, mode="RGBA")).convert("RGB")


def run_depthpolyp(image: Image.Image, threshold: float):
    if image is None:
        raise gr.Error("Please upload an image first.")

    session, input_name = get_session()
    original, original_size, tensor = preprocess(image)
    segmentation, depth = session.run(None, {input_name: tensor})

    seg_prob = segmentation[0, 0]
    depth_prob = depth[0, 0]

    seg_image = to_grayscale(seg_prob, original_size)
    mask = seg_image.point(lambda value: 255 if value >= int(threshold * 255) else 0)
    depth_image = colorize_purple_yellow(depth_prob, original_size)
    overlay = make_overlay(original, seg_image)

    return overlay, mask, depth_image


def example_paths():
    if not EXAMPLES_DIR.exists():
        return []
    return [[str(path), 0.3] for path in sorted(EXAMPLES_DIR.glob("*")) if path.suffix.lower() in {".jpg", ".jpeg", ".png"}]


with gr.Blocks(title="DepthPolyp Demo") as demo:
    gr.Markdown(
        """
        # DepthPolyp
        CPU ONNX demo for pseudo-depth guided colonoscopic polyp segmentation.
        """
    )

    with gr.Row():
        with gr.Column(scale=1):
            input_image = gr.Image(type="pil", label="Colonoscopy image")
            threshold = gr.Slider(0.05, 0.95, value=0.3, step=0.05, label="Mask threshold")
            run_button = gr.Button("Run inference", variant="primary")
        with gr.Column(scale=2):
            with gr.Row():
                overlay = gr.Image(type="pil", label="Segmentation overlay")
                mask = gr.Image(type="pil", label="Binary mask")
            depth = gr.Image(type="pil", label="Pseudo-depth")

    run_button.click(
        fn=run_depthpolyp,
        inputs=[input_image, threshold],
        outputs=[overlay, mask, depth],
    )
    input_image.change(
        fn=run_depthpolyp,
        inputs=[input_image, threshold],
        outputs=[overlay, mask, depth],
    )
    threshold.change(
        fn=run_depthpolyp,
        inputs=[input_image, threshold],
        outputs=[overlay, mask, depth],
    )

    gr.Examples(
        examples=example_paths(),
        inputs=[input_image, threshold],
        outputs=[overlay, mask, depth],
        fn=run_depthpolyp,
        cache_examples=False,
    )

    gr.Markdown(
        "Model: [ReaganWZY/DepthPolyp](https://huggingface.co/ReaganWZY/DepthPolyp). "
        "For research use only; not for clinical diagnosis."
    )


if __name__ == "__main__":
    demo.launch(server_name="0.0.0.0", server_port=7860)