Local binary descriptor constitutes power visual cues for feature representation. They provide discriminative information about small appearance details in local neighbourhoods. So, they are robust to local changes databases such as illumination, identity, and expression. Unlike existing local descriptors is not discriminatory enough to estimate the relationship between two people. This is mainly due to the learning feature code individually and the hand-crafted features which previous knowledge is required. In this paper, we propose an effective Context-Aware Local Binary Feature Learning (CA-LBFL)for kinship verification in order to solve the proposed problem. (CA-LBFL)a method has applied to learn contextual features from raw pixels directly and to eliminates the dependence on hand-crafted features. Experimental results demonstrate that the proposed method achieves competitive results compared with other states-of-the-art.