motorcycle-license-plate-detection

Motorcycle License Plate Detector

YOLOv8n 기반 이륜차 번호판 검출 모델입니다. CCTV 또는 일반 이미지에서 이륜차 번호판 영역을 bounding box로 검출하는 용도로 파인튜닝되었습니다.

이 모델 저장소는 공개 가능한 최종 검출 모델 가중치와 모델 카드만 제공합니다. 학습 데이터, 검증 이미지, 라벨 파일, TensorBoard/MLflow 로그는 포함하지 않습니다.

Model

Metrics

Usage

Install dependencies:

pip install ultralytics huggingface_hub

Download and run inference:

from huggingface_hub import hf_hub_download
from ultralytics import YOLO

repo_id = "gkes-lab/motorcycle-license-plate-detection"

model_path = hf_hub_download(
    repo_id=repo_id,
    filename="yolov8_best_plate_detector.pt",
)

model = YOLO(model_path)
results = model("path/to/image.jpg", conf=0.5)

for result in results:
    for box in result.boxes:
        print(box.xyxy, box.conf)

Training Code

Training script and reproducible configuration are available on GitHub:

https://github.com/gkes-lab/motorcycle-license-plate-detection

Data Notice

The training and validation images are not included in this repository because they may contain sensitive CCTV and license plate information. Only model weights and technical metadata are released.

Intended Use

This model is intended for research, demonstration, and smart-city safety applications involving motorcycle license plate detection.

Users are responsible for complying with applicable privacy, surveillance, traffic, and data protection laws when applying this model to real-world images or video streams.

Limitations

• Performance may vary depending on camera angle, motion blur, low-light scenes, weather, occlusion, and image resolution.
• This repository does not include end-to-end violation detection or administrative enforcement logic.

License

Apache License 2.0