app.py

import gradio as gr
import numpy as np
import json
import os
from PIL import Image
import onnxruntime as rt

class ONNXModel:
    def __init__(self, dir_path) -> None:
        """Load model metadata and initialize ONNX session."""
        model_dir = os.path.dirname(dir_path)
        with open(os.path.join(model_dir, "signature.json"), "r") as f:
            self.signature = json.load(f)

        self.model_file = os.path.join(model_dir, self.signature.get("filename"))
        if not os.path.isfile(self.model_file):
            raise FileNotFoundError("Model file does not exist.")

        self.signature_inputs = self.signature.get("inputs")
        self.signature_outputs = self.signature.get("outputs")

        if "Image" not in self.signature_inputs:
            raise ValueError("ONNX model must have an 'Image' input. Check signature.json.")

        # Check export version
        version = self.signature.get("export_model_version")
        if version is None or version != EXPORT_MODEL_VERSION:
            print(f"Warning: Expected model version {EXPORT_MODEL_VERSION}, but found {version}.")

        self.session = None

    def load(self) -> None:
        """Load the ONNX model with execution providers."""
        self.session = rt.InferenceSession(self.model_file, providers=["CPUExecutionProvider"])

    def predict(self, image: Image.Image) -> dict:
        """Process image and run ONNX model inference."""
        img = self.process_image(image, self.signature_inputs["Image"]["shape"])
        feed = {self.signature_inputs["Image"]["name"]: [img]}
        output_names = [self.signature_outputs[key]["name"] for key in self.signature_outputs]
        outputs = self.session.run(output_names=output_names, input_feed=feed)
        return self.process_output(outputs)

    def process_image(self, image: Image.Image, input_shape: list) -> np.ndarray:
        """Resize and normalize the image."""
        width, height = image.size
        if image.mode != "RGB":
            image = image.convert("RGB")

        square_size = min(width, height)
        left = (width - square_size) / 2
        top = (height - square_size) / 2
        right = (width + square_size) / 2
        bottom = (height + square_size) / 2
        image = image.crop((left, top, right, bottom))

        input_width, input_height = input_shape[1:3]
        image = image.resize((input_width, input_height))
        image = np.asarray(image) / 255.0
        return image.astype(np.float32)

    def process_output(self, outputs: list) -> dict:
        """Format the model output."""
        out_keys = ["label", "confidence"]
        results = {key: outputs[i].tolist()[0] for i, key in enumerate(self.signature_outputs)}
        confs = results["Confidences"]
        labels = self.signature["classes"]["Label"]
        output = [dict(zip(out_keys, group)) for group in zip(labels, confs)]
        return {"predictions": sorted(output, key=lambda x: x["confidence"], reverse=True)}

EXPORT_MODEL_VERSION = 1
model = ONNXModel(dir_path="model.onnx")
model.load()

def predict(image):
    """Run inference on the given image."""
    image = Image.fromarray(np.uint8(image), "RGB")
    prediction = model.predict(image)
    for output in prediction["predictions"]:
        output["confidence"] = round(output["confidence"], 4)
    return prediction

inputs = gr.Image(image_mode="RGB")
outputs = gr.JSON()

description = (
    "This is a web interface for the Naked Detector model. "
    "Upload an image and get predictions for the presence of nudity.\n"
    "Model and website created by KUO SUKO, C110156115 NKUST."
)

interface = gr.Interface(
    fn=predict,
    inputs=inputs,
    outputs=outputs,
    title="Naked Detector",
    description=description
)

interface.launch()

# this is changed by ChatGPT, if it run like a shit. don't blame me ><