Identifying age, gender, and ethnicity of a person is still a difficult undertaking, given the absence of a specific definition of ethnicity, the wide variety of age distributions and the various aging factors which primarily affects the facial traints. Therefore, this paper propose four deep learning models which extract and classify the age, gender, ethnicity and the entire set of demographic attributes at the same time as well as two data sampling methods to deal with the imbalanced nature of the chosen dataset. The accuracy for the proposed gender, ethnicity and the multi-label models were 96.14%, 88.72% and 93.09% respectively and were trained and tested using the UTK-Face dataset.