Parental behavior is widely preserved in living animals. A caregiving of mother is needed for behavioral and emotional development of child. Various responsive hormones have been identified to play the important roles on maternal behavior. One of neurohypophyseal hormone, arginine vasopressin, has been known to regulate maternal behavior mainly through activation of vasopressin V1a subtype receptors in the central nervous system. Moreover, administration of antagonist of another vasopressin receptor subtype, V1b receptor, into lateral ventricle changed maternal care. However, the knowledge of V1b receptors on maternal behavior is limited. Recently, machine learning algorithms have been applied in several scientific studies, in which the computer vision and deep learning are a brunch of this intelligence algorithm. Taking the advantage of computer science into account, we had employed this method to investigate a possible role of vasopressin in maternal behavior in mother and pup interaction during lactation period. In each image from video recording, one or more objects can be analyzed and classified by deep learning models. We found that intimal relationship between dam and pups can be analyzed in detail. Furthermore, our method can be applied to a large scale of dataset. In summary, we have successfully developed deep learning models to study the relationships between dam and pups of wildtype and gene knockout mice in free moving conditions. Furthermore, our developed method effectively analyzes the complex behavior of laboratory animals with increasing efficiency and consistency.