MS-Celeb-1M

by Yandong Guo,Lei Zhang,Yuxiao Hu,Xiaodong He,Jianfeng GaoUnknown

MS-Celeb-1M

To facilitate the above face recognition task, we provide a large training dataset which covers the top 100K celebrities. This training dataset is prepared by the following steps. First, we select the top 100K entities from the 1M celebrity list in terms of their popularities. Then, we leverage public search engines to provide approximately 100 images for each celebrity, resulting in about 10M web images. Note that the dataset is mainly to facilitate the participants to quickly get started. In the contest, we do not limit the use of external data, but encourage the participants to treat data collection as part of the face recognition challenge.

Dataset Attributes

Label SVG
TasksFacial Recognition, Classification
Label SVG
CategoriesCelebrities, Pop Stars, Actors
Label SVG
SensorRGB Camera