johnamit/motoreid

MotoReID provides the trained model assets used for MotoGP team detection and team classification from race broadcast footage. The full project combines YOLOv8 for motorcycle detection, DINOv3 for visual feature extraction, a lightweight logistic regression classifier for team prediction, and BoT-SORT configuration for tracked inference in the main pipeline.

Repository Contents

This Hugging Face repository stores the model assets used by the main MotoReID GitHub project.

Layout

Path	Description
assets/MotoReID.png	Project banner image
models/DINO/	DINOv3 backbone weights used for feature extraction
models/classifier/	Trained team-classification model
models/YOLO/	Fine-tuned YOLO detector weights and tracker configuration

Pipeline Overview

Video Frame -> YOLOv8 Detection -> Crop Extraction -> DINOv3 Embedding -> Team Classification
For tracked demo/inference runs, the main project also uses BoT-SORT configuration stored in the YOLO model directory.

Supported Teams (2025 Season)

Manufacturer	Teams
Ducati	Lenovo Team, Gresini Racing, VR46 Racing
Aprilia	Factory Racing, Trackhouse Racing
KTM	Factory Racing, Tech3
Honda	Repsol HRC, LCR Honda
Yamaha	Monster Energy, Prima Pramac

How To Use

Option 1: Download the full repository

Clone or download this Hugging Face repo, then place the models/ folder into your local project root so the code can find the weights using the expected paths.

Expected local structure:

MotoGP-Team-Detection/
├── assets/
├── data/
├── models/
│   ├── DINO/
│   ├── YOLO/
│   └── classifier/
├── results/
├── src/
│   ├── data/
│   ├── inference/
│   ├── scripts/
│   └── training/
├── README.md
└── requirements.txt

Option 2: Download with huggingface_hub

from huggingface_hub import snapshot_download

snapshot_download(
    repo_id="johnamit/motoreid",
    local_dir="motoreid_hf_repo"
)

Project Scope

This repository contains model assets only.

The following are hosted separately:

• source code and documentation: GitHub
• large raw data and derived outputs: Google Drive

Notes

• This repo does not include the training dataset.
• These assets are intended to be used with the code in the main GitHub repository.
• Folder names are preserved to match the paths expected by the project code.

License

MIT License

Acknowledgements

• Ultralytics (https://ultralytics.com/) for YOLOv8
• Meta AI (https://github.com/facebookresearch/dinov3) for DINOv3
• Roboflow (https://roboflow.com/) for annotation tooling