The VGGFace2 dataset proposed by Cao et al. is annotated with 9,131 unique people with 3.31 million images. The dataset can be obtained from http://www.robots.ox.ac.uk/~vgg/data/vgg_face2/. The variation includes age, ethnicity, pose, profession, and illumination. This is the largest dataset available for face verification.
Here is a sample of the images present in the dataset:
Proposed by Cao et al. and reproduced from http://www.robots.ox.ac.uk/~vgg/data/vgg_face2/web_page_img.png
The minimum, mean, and maximum number of images per unique person are 87, 362.6, and 843 respectively.