Biometric systems are mostly preferred by different fields because of their higher security features. In past, passwords were used to identify and validate the genuine users. However, it was not preferred because these can be easily accessed by the third party. Hence, consequently it increases the frequency of fraudulent and breach incidents. In contrast to this, biometric systems provide full authentication of an individual based on their physiological traits and modalities that are used as input to biometric system. The paper presents a review that is mainly focused upon the four physiological modalities; face recognition, fingerprints recognition, iris recognition, and retina scan. In the knowledge of existing challenges the paper illustrates the different studies presented by research community in the similar field. A comparative analysis of physiological modalities is presented in the paper that shows the rising popularity of deep learning techniques in the biometric design.