Head and face shape classification and type prediction of female college students based on MLP neural network

ABSTRACT

Currently with the changes in living habits and eating habits of China consumers have higher requirements for the wearing comfort and fit of head and face products such as helmets and masks. In addition the outbreak of COVID-19 in 2019 has made suitable masks an important protective equipment for medical staff and the general population. How to improve the safety protection level of masks has also become a hot social issue of concern. The fit of the mask is directly related to the protection effectiveness so it is urgent to measure track and update human head and face data. The research on the characteristics and classification of human head and face is an important basis for the structural design size formulation fit research and plate shape optimization of masks and helmets. Multilayer perceptron is an ANN algorithm. With the development of neural network technology it is gradually applied to prediction and classification. The model with strong nonlinear approximation function simple structure controllable number of input variables and strong operability can be applied to the classification and prediction of human body shape. In order to improve the adaptability of head and face products this paper took 189 female college students aged 18 - 26 as the research subjects and used the Martin measuring instrument to measure the head and face of the subjects. Feature factors affecting head and face shape were extracted by principal component analysis PCA the K-Means method was used to classify the head and face morphology and the index classification method was used to quantify head and face morphology. As a result a head-face shape prediction model based on MLP-ANN was proposed to improve the problem of low production work efficiency caused by too many head and face sizes in classifying or selecting models with too many references. The study found that through the analysis of head and face characteristics of 189 subjects seven important characteristic factors affecting the head and face shape were extracted head contour factor morphological facial factor morphological facial factor eye factor nose factor and mouth & lip factor. The head and face shapes were divided into five sizes according to the clustering center value of each category XS type/morphological index > 93 S type/morphological index 88 93 M type/morphological index 84 88 L type/morphological index 79 84 XL type/morphological index 79 and the M type was the most widely distributed and had a big coverage rate so it can be used as an intermediate type. Then through the MLP neural network seven head-face feature factors were used to predict head-face shape classification. The generated model had a 93. 42% correct prediction result and the research results can provide a reference for the design and production of head and face products. This paper provides an objective method for the study of head and facial features but there are still some limitations. In the future we can continue to improve the classification of head and face shape by expanding the area and age of the experimental subjects for comparative research. We can apply the classification to the head and face product specification system so as to accumulate morphological data for the study of the head and face characteristics of contemporary Chinese people and the design of head and face products such as masks for the Chinese market. © 2022 China Silk Association. All rights reserved.