| from __future__ import annotations |
|
|
| import io |
| from pathlib import Path |
|
|
| from PIL import Image |
|
|
| from .config import DEFAULT_BACKEND, DEFAULT_HF_MODEL_ID, DEFAULT_HYBRID_UNIVFD_WEIGHT |
| from .model import create_detector |
|
|
|
|
| def create_app( |
| *, |
| threshold: float = 0.5, |
| device: str | None = None, |
| backend: str = DEFAULT_BACKEND, |
| weight_path: str | Path | None = None, |
| hf_model: str = DEFAULT_HF_MODEL_ID, |
| hybrid_univfd_weight: float = DEFAULT_HYBRID_UNIVFD_WEIGHT, |
| ): |
| try: |
| from fastapi import FastAPI, File, UploadFile |
| except ImportError as exc: |
| raise RuntimeError("Missing dependency: install ai-image-detector[api].") from exc |
|
|
| detector = create_detector( |
| backend, |
| device=device, |
| threshold=threshold, |
| weight_path=weight_path, |
| hf_model=hf_model, |
| hybrid_univfd_weight=hybrid_univfd_weight, |
| ) |
| app = FastAPI(title="AI Image Detector", version="0.3.0") |
|
|
| @app.get("/health") |
| def health() -> dict: |
| return {"status": "ok", "model": detector.model_info()} |
|
|
| @app.post("/detect") |
| async def detect(file: UploadFile = File(...)) -> dict: |
| content = await file.read() |
| with Image.open(io.BytesIO(content)) as image: |
| result = detector.predict_image(image) |
| payload = result.as_dict() |
| payload["filename"] = file.filename |
| return payload |
|
|
| return app |
|
|
