MS-Celeb-1M

Challenge of Recognizing One Million Celebrities in the Real World

Learn More ...

MS-Celeb-1M

We provide training and benchmark testing dataset for the following task: recognizing one million celebrities from their face images and link them to the corresponding entity keys in a knowledge base. More specifically, we provide,

  • MS-Celeb-1M Training v1: about 10M images for 100K celebrities
  • Concrete measurement to evaluate the performance of recognizing one million celebrities
  • Lowshot learning setup and testbed

News

  • 03/16/2018 More challenges will be hosted this year!
  • 03/16/2018 Azure blob download link updated!

Citation

If you are reporting results of the challenge or using the dataset, please cite the paper "MS-Celeb-1M: A Dataset and Benchmark for Large Scale Face Recognition".

        @INPROCEEDINGS { guo2016msceleb,
            author = {Guo, Yandong and Zhang, Lei and Hu, Yuxiao and He, Xiaodong and Gao, Jianfeng},
            title = {M{S}-{C}eleb-1{M}: A Dataset and Benchmark for Large Scale Face Recognition},
            booktitle = {European Conference on Computer Vision},
            year = {2016},
            organization={Springer}}

A paper defines the low-shot face recognition benchmark, with a baseline method: "One-shot Face Recognition by Promoting Underrepresented Classes".

        @article { lowshotface,
            author = {Guo, Yandong and Zhang, Lei},
            title = {One-shot Face Recognition by Promoting Underrepresented Classes},
            Journal   = {arXiv preprint arXiv:1707.05574},
            Year      = {2017}}