A Systematic Mapping on Detection of Human Mouth Landmarks

Abstract

Facial landmarks represent regions of interest whose detection and localization generate features supporting the identification of movements, feelings, and reactions. Most facial feature detection algorithms focus on entire semantic areas, such as the region of a mouth which allows grained manipulation that is essential for a wide domain variety. This paper describes a systematic mapping of the detection of landmarks in human faces and their application domains. The identification and selection methods of primary studies include automatic search on information sources, inclusion, and exclusion criteria over 344 scientific papers from 2015 and 2021, from which we analyzed and synthesized 115 primary studies. Our analysis considered the implementation of methods, types, and uses of data extracted from the mouth. The mapping brought exciting information as new methods, datasets, and domains researched through the time interval reviewed as well research gaps that can be explored.