Extracting Multi-Dimensional Features for BMI Estimation Using a Multiplex Network

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Body Mass Index (BMI) is a crucial indicator for assessing human obesity and overall health, providing valuable insights for applications such as health monitoring, patient re-identification, and personalized healthcare. Recently, several data-driven methods have been developed to estimate BMI using 2D and 3D features extracted from facial and body images or RGB-D data. However, current research faces challenges such as the incomplete consideration of anthropometric features, the neglect of multiplex networks, and low-BMI-estimation performance. To address these issues, this paper proposes three 3D anthropometric features, one 2D anthropometric feature, and a deep feature extraction method to comprehensively consider anthropometric features. Additionally, a BMI estimation method based on a multiplex network is introduced. In this method, three types of features are extracted by constructing a multichannel network, and BMI estimation is performed using Kernel Ridge Regression (KRR). The experimental results demonstrate that the proposed method significantly outperforms state-of-the-art methods. By incorporating symmetry into our analysis, we can uncover deeper patterns and relationships within complex systems, leading to a more comprehensive understanding of the phenomena under investigation.
Funding Information
  • National Key Research and Development Program of China (2022YFC3502302, 82204770, KYCX23_2078)
  • National Youth Science Foundation of China (2022YFC3502302, 82204770, KYCX23_2078)
  • Graduate Research Innovation Program of Jiangsu Province (2022YFC3502302, 82204770, KYCX23_2078)

This publication has 50 references indexed in Scilit: