AN ALGORITHM FOR GENERATING DESCRIPTIVE SENTENCES OF THE HUMAN HEAD PARTS BASED ON ENGLISH GRAMMAR

Abstract

Human head exhibits many biological features (attributes) that represent the characteristics of the human head with robust inherent stability and individual variation. These attributes provide important discriminative knowledge about humans, such as gender, age, race, hairstyle, hair color, etc. Recently, several human head attribute classification networks have been proposed. However, these networks do not provide a clear picture of the human head because they predict head attributes in terms of binary values (i.e., 0 or 1) or by their labels (i.e., male, young). Therefore, in this study, a description algorithm was proposed to describe the main characteristics of the human head using the adjective’s arrangement rules. The proposed algorithm was reviewed by experts, and the responses of seven experts show that the algorithm follows the adjective’s arrangement rules in accordance with the conventions of human language. The experts also found the descriptive sentences acceptable, understandable, and grammatically correct.