Skip to main content

ORIGINAL RESEARCH article

Front. Artif. Intell.
Sec. Pattern Recognition
Volume 7 - 2024 | doi: 10.3389/frai.2024.1336320

Using Synthetic Data for Semantic Segmentation of the Human Body in the Problem of Extracting Anthropometric Data

Provisionally accepted
Azat Absadyk Azat Absadyk *Olzhas Turar Olzhas Turar Darkhan Akhmed-Zaki Darkhan Akhmed-Zaki
  • Astana IT University, Astana, Kazakhstan

The final, formatted version of the article will be published soon.

    The COVID-19 pandemic has highlighted the lack of preparedness in various sectors, including global entrepreneurship, for remote work. Clothing sellers, in particular, face the challenge of virtualizing their products without the ability for customers to try them on, necessitating an easy and accurate method to obtain anthropometric data to reduce returns, logistics costs, and environmental waste. To obtain anthropometric data, it is necessary to find key points from a person's image that are located around the perimeter of the silhouette. Many trained neural networks can determine the silhouette of a person from an image today. The main problem is the lack of a dataset, which determines only the shape of the human body without consideration of the shape of clothes. In order to obtain anthropometric data, it is necessary to find key points considering human clothing. The algorithm must be able to determine the true body shape.Obtaining such type of data is almost impossible. Using the NVIDIA Omniverse Replicator tool, we created a synthetic dataset comprising more than 7000 input and output images. We tested the resulting set for training popular architectures for semantic segmentation with popular CNN architectures as a backbone. The results show that this set can be applied to real images.

    Keywords: synthetic data, Human Segmentation, Anthropometry, CNN, Nvidia replicator, humanbody

    Received: 10 Nov 2023; Accepted: 28 Jun 2024.

    Copyright: © 2024 Absadyk, Turar and Akhmed-Zaki. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

    * Correspondence: Azat Absadyk, Astana IT University, Astana, Kazakhstan

    Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.