Multi-view Deep Features for Robust Facial Kinship Verification

摘要: Automatic kinship verification from facial images is an emerging research topic in machine learning community. In this paper, we proposed effective features extraction model based on multi-view deep features. Thus, used four pre-trained models using eight layers (FC6 and FC7 of each VGG-F, VGG-M, VGG-S VGG-Face models) to train the Multilinear Side-Information Discriminant Analysis integrating Within Class Covariance Normalization (MSIDA+WCCN) method. Furthermore, show that how can metric methods WCCN method integration improves Simple Scoring Cosine similarity (SSC) We refer SSC RFIW'20 competition concatenation. decreases intra-class variations effect introduced by weights. evaluate our two benchmarks namely KinFaceW-I KinFaceW-II databases Parent-Child relations (Father-Son, Father-Daughter, Mother-Son Mother-Daughter). MSIDA+WCCN with 12.80% 14.65% databases, respectively. The results obtained are positively compared some modern methods, including those rely learning.