Please see https://github.com/yakhyo/face-recognition to train Face Recognition model.

MS1M-ArcFace Dataset Description

The MS1M-ArcFace dataset is a cleaned and refined version of the original MS-Celeb-1M dataset, specifically curated for face recognition tasks. This dataset was processed to remove noisy and misaligned images, improving its quality and usability in training robust face recognition models.

Features - Image Size: 112x112 pixels - Classes: 85742 - Aligned: Standardized facial landmarks

Info - Dataset Origin: Based on the MS-Celeb-1M dataset, originally released by Microsoft - Research. - Purpose: Designed to facilitate research and development in face recognition, particularly for high-accuracy models. - Data: Contains millions of images of celebrity faces, preprocessed and aligned for optimal model training. - Preprocessing: Cleaned and refined using advanced methods to reduce noise, mislabels, and inaccuracies. - Applications: Used in training state-of-the-art models like ArcFace for tasks such as identity verification, facial feature extraction, and more. - License: Users should verify compliance with ethical and licensing requirements before using or distributing the dataset. - This dataset has been extensively used in academic and industrial research for benchmarking and developing cutting-edge face recognition systems.

Please refer to the original source of this dataset for additional information. It's released here for academic purposes only.

In this paper, we design a benchmark task and provide the associated datasets for recognizing face images and link them to corresponding entity keys in a knowledge base. More specifically, we propose a benchmark task to recognize one million celebrities from their face images, by using all the possibly collected face images of this individual on the web as training data. The rich information provided by the knowledge base helps to conduct disambiguation and improve the recognition accuracy, and contributes to various real-world applications, such as image captioning and news video analysis. Associated with this task, we design and provide concrete measurement set, evaluation protocol, as well as training data. We also present in details our experiment setup and report promising baseline results. Our benchmark task could lead to one of the largest classification problems in computer vision. To the best of our knowledge, our training dataset, which contains 10M images in version 1, is th

This is MS1M-refine-v2 (a.k.a. MS1M-ArcFace) dataset for facial recognition in TFRecord dataset format. It is from InsightFace DatasetZoo.

Dataset

This dataset was created by emIkhlas

Contents

Glint360K contains ** 17091657 ** images of ** 360232 ** individuals. By employing the Patial FC training strategy, baseline models trained on Glint360K can easily achieve state-of-the-art performance. Detailed evaluation results on the large-scale test set (e.g. IFRT, IJB-C and Megaface) are as follows: # 1. Evaluation on IFRT ** r ** denotes the sampling rate of negative class centers. | Backbone | Dataset | African | Caucasian | Indian | Asian | ALL | | —————— | —————- | ——- | ——- | ——— | ——- | ——- | | R50 | MS1M-V3 | 76.24 | 86.21 | 84.44 | 37.43 | 71.02 | | R124 | MS1M-V3 | 81.08 | 89.06 | 87.53 | 38.40 | 74.76 |