AI-CLS Service

FastAPI 微服务，基于 YOLOv8 识别图片中的校园常见失物并返回前三个标签。分类集合是从 COCO-80 精选并重命名的英文标签，聚焦校园场景。

当前默认的标签清单：

• backpack
• umbrella
• shoulder bag or wallet
• luggage or suitcase
• ball or sports gear
• water bottle or thermos
• tumbler
• cup
• laptop
• mouse
• presentation clicker or remote
• keyboard
• phone or tablet
• notebook or campus card
• watch or desk clock
• stationery (scissors-like)
• plush toy
• toiletry (toothbrush)

Getting Started

1. Create a virtual environment

   python3 -m venv .venv
   source .venv/bin/activate

2. Install dependencies

   pip install -r requirements.txt

3. Download the YOLOv8s weights

   mkdir -p models
   curl -L -o models/yolov8s.pt https://github.com/ultralytics/assets/releases/download/v0.0.0/yolov8s.pt

4. Start the service

   uvicorn main:app --host 0.0.0.0 --port 5101

   You can also run python main.py to launch Uvicorn directly.

API Overview

• GET /health – Returns { "status": "ok" }.
• POST /predict-category – Accepts a single image file and returns its top-3 labels.
• POST /predict-category-batch – Accepts 1–3 image files and returns the top-3 labels for each image.

All responses return the English COCO-80 labels alongside the normalised confidences.