Social Networks Faces (SNFA) database is a face dataset of social networks, which reflects the real world combination of shared images from social networks. The database contains multiple types of facial images, including natural faces, faces with pet decoration, cosplay's faces and masked faces, which can be deemed the largest combinations of different types of facial images in one dataset itself. It is used for scientific research of face detection. The face images in the database are crawled from the Internet by Fujian Key Laboratory of Integrated Circuit Design and Measurement (Xiamen University) for educational and non-commercial use. The images in the database may be subject to copyright, we do not make them publicly available on our site.

Details

-The images are collected from social network Weibo¹ and Instagram² website.

-There are 13,236 photos in the dataset.

-There is at least one face in each picture.

-The images are kept with a minimal area of 300*300 pixels.

-The proportions of natural faces, faces with pet decoration, cosplay's faces and masked faces are 92.6%, 3.7%, 2.1%,1.6% respectively.

-There are 56 photos are used in our paper "Face Detection of Social Networks based on Orthogonalized and Modified Multi-task Cascaded Convolutional Networks" for test experiments, which are publicly available.

Samples

1 https://weibo.com

2 https://www.instagram.com