Fine-grained LFW database

Abstract

Current deep learning methods have achieved human-level performance on Labeled Faces in the Wild (LFW) database, but we think it is because that the limited number of pairs on LFW do not capture the real difficulty of large-scale unconstrained face verification problem. Besides the intra-class variations like pose, illumination, occlusion and expression, highly visually similarity of different persons' faces is an another challenge. It is unavoidable in large dataset and many researchers ignore it. Therefore, in this paper, we firstly select some visually similar pairs in LFW database by combining the deep learning method and human annotation results. Preserving the matched pairs and replacing the mismatched pairs of LFW with the selected similar pairs, we obtain the Fine-grained LFW (FGLFW) database which can better reflect the real difficulty of face verification. Experimental results show that methods achieving not bad performance on LFW drops more than 11% even 25% on FGLFW. It reflects that visually similar pairs are difficult to current methods and our FGLFW database is a quite challenging database. Researchers still have a long way to go for solving face verification problem on such a database.