Race-Classification-FairFace-YOLOv8 Open Source Model - Quickly Complete Facial Race Classification Tasks

Race Classification FairFace YOLOv8

Developed by Anzhc

A fast race classification model based on YOLOv8/YOLO11 architecture, trained on the FairFace dataset for facial race classification tasks.

Face-related

PyTorch

#Fast Race Classification #Low Computational Cost #Facial Attribute Analysis

Downloads 100

Release Time : 2/8/2025

Model Overview

This model focuses on race classification of facial images, supporting 7 race categories (Black, East Asian, Indian, Latino/Hispanic, Middle Eastern, Southeast Asian, White), aiming to reduce classification bias through a balanced dataset.

Model Features

Fast Inference

Designed for rapid labeling, suitable for applications requiring efficient processing of large volumes of images.

Low-bias Classification

Trained on the balanced FairFace dataset to mitigate bias issues in race classification.

Multiple Model Options

Offers various model variants from lightweight (v8n) to high-performance (11x) to meet different performance needs.

Model Capabilities

Facial Image Classification

Race Attribute Recognition

Batch Image Processing

Use Cases

Demographic Analysis

Social Media Analysis

Analyze the racial distribution of user-uploaded photos for demographic research.

Content Moderation

Diversity Monitoring

Monitor the racial diversity distribution of platform content.

🚀 Race Classification YOLOv8/11

This model is designed for race classification, leveraging the YOLOv8/11 architecture to provide fast and efficient tagging capabilities.

🚀 Quick Start

This model is based on the FairFace 0.25 padding variant dataset composed by Microsoft researchers. The goal is to reduce bias by better balancing classes in the dataset.

Karkkainen, Kimmo, and Joo, Jungseock.
FairFace: Face Attribute Dataset for Balanced Race, Gender, and Age for Bias Measurement and Mitigation.
Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2021, pp. 1548–1558.
FairFace Dataset on GitHub

You can also find their pretrained model here.

This YOLOv8 training is specifically for race classification. The author aimed for a very fast model for tagging, and this model is likely useful for such purposes.

There are simple scripts available for you to use on your data. By default, it will output .txt files (or append to existing ones), so you can modify them according to your specific needs: Simple Utility Scripts for YOLO

📚 Documentation

Model Performance

Model	Target	top1_acc	Classes	Dataset size	Training Resolution
Race-CLS-FairFace_yolov8n	Face: Real	0.717	7 (Black, East Asian, Indian, Latino_Hispanic, Middle Eastern, Southeast Asian, White)	~86740 (train), ~10950 (val)	224
Race-CLS-FairFace_yolov8s	Face: Real	0.721	7 (Black, East Asian, Indian, Latino_Hispanic, Middle Eastern, Southeast Asian, White)	~86740 (train), ~10950 (val)	224
Race-CLS-FairFace_yolov8m	Face: Real	0.725	7 (Black, East Asian, Indian, Latino_Hispanic, Middle Eastern, Southeast Asian, White)	~86740 (train), ~10950 (val)	224
Race-CLS-FairFace_yolo11l	Face: Real	0.733	7 (Black, East Asian, Indian, Latino_Hispanic, Middle Eastern, Southeast Asian, White)	~86740 (train), ~10950 (val)	224
Race-CLS-FairFace_yolo11x	Face: Real	0.735	7 (Black, East Asian, Indian, Latino_Hispanic, Middle Eastern, Southeast Asian, White)	~86740 (train), ~10950 (val)	224

📄 License

This project is licensed under the AGPL-3.0 license.

📦 Model Information

Property	Details
Model Type	YOLOv8/11 for race classification
Training Data	FairFace 0.25 padding variant dataset
Library Name	ultralytics
Pipeline Tag	image-classification
Base Model	Ultralytics/YOLOv8, Ultralytics/YOLO11
Tags	pytorch, YOLOv8, Ultralytics, YOLO11

Model Index

Model Name	Task	Dataset	Metrics
v8n	Image Classification	FairFace	top1_acc: 0.717
v8s	Image Classification	FairFace	top1_acc: 0.721
v8m	Image Classification	FairFace	top1_acc: 0.725
11l	Image Classification	FairFace	top1_acc: 0.733
11x	Image Classification	FairFace	top1_acc: 0.735

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご