The pattern of feelings and actions of a human being is personality. From ancient times, we have been interested in knowing the personalities of others and acting accordingly. Although determining actual personality of a human is a complex task and it is like uncovering layers of an onion, perception of personality is relatively easy. We can look at a person’s face, and within seconds, we can judge many aspects, like gender or age. Additionally, we can make decisions about whether we like a person or not based on how friendly that person is looking. In this paper, detection of apparent personality with transfer learning models is experimented with very small amount of data in the form of images of people talking in front of camera. Three customized state-of-the-art models have been trained for regression and achieved a decent accuracy below 60% mean absolute error.