Automatic Identification of Somatotype by Digital Images
Keywords: Somatotype, Feature Selection, Artificial Neural Network, Genetic Algorithm
AbstractSomatotype is a metric that concerns human body shape and composition. It is important in many applications, especially for the fields of physical education and health. However, obtaining the somatotype is very time-consuming, it requires an expert of the area to take several measurements manually on the individual's body, demanding several anthropometric devices, some of which are not easily portable. The proposal of this work is to obtain the somatotype of bodybuilders using their body images in different positions, based on image processing and machine learning techniques. Thus, a dataset of 46 bodybuilders was obtained along the years of 2014 to 2016 in the state of Espirito Santo. The results obtained show that the somatotype of bodybuilders can be estimated reasonably based only on their images, obtaining the best classification rate of 92%, which is an inexpensive alternative to obtain this information to actual days.