- Project Duration: 01/29/2024-12/19/2024

- InBody Co., Ltd (Internship)

- Consumer Electronics Show 2025

- Team Member: Jaeho Kim (Team Lead.), Sewon Kim (SW & AI Lead.), Seungseo Park, Soohyeon Ga (Hardware Lead.), Yeongjin Kim, Wongoon Lee, Mingwan Seol, Soohong Lim

As the SW & Vision AI Team Leader, I have developed a computer vision AI solution for the new InBody Scale AI product. This involved researching and enhancing the performance of height estimation algorithms using a low-performance monocular camera and limited computing resources. Additionally, I developed a face verification solution, along with the necessary pipeline and database for it.

Existing height estimation algorithms and papers using vision assume that the user's entire body is within the camera's field of view. However, due to the mechanical characteristics of the InBody device, the user's full body does not fit within the camera's field of view. Our attempt to estimate height under these conditions is unprecedented. In our first experiment, we achieved an average error of 0.6% and a maximum error of 1%. We tackled the ultimate problem of height estimation by breaking it down into sub-problems where computer vision AI can excel: Object Detection and Semantic Segmentation. To achieve this, we designed and utilized a structure that allows the camera to move via motors and rails.

InBody AI scale reduces the speed of existing body composition analyzers by less than half and automatically measures and inputs all information required for body composition analysis. This allows users to conveniently measure as if they were stepping on a scale, and makes body composition measurement a natural part of daily life. It is the world's first all-in-one body composition analyzer that maximizes user convenience in the measurement process through AI and an improved electrode system, and also provides detailed comparison values and AI-tailored health services through big data.

The InBody AI scale is the world's first AI body composition analyzer to completely eliminate the inconvenience of manual input by automatically entering all essential information through AI. The AI camera moving through the vertical linear structure is equipped with the self- developed Eye 2 Height Algorithm(E2H Algorithm) based on HOG detector model and CNN predictor model, and it measures the eye level within 0.2 inches of error through more than 3,000 user data and automatically calculates the height. The AI face recognition implemented as a deep-face model automatically inputs user information and completes the measurement within 30 seconds without stopping. Afterward, detailed age-based comparisons and AI-customized health information services are provided using 100 million data points stored in the InBody cloud. Since body composition analysis is performed by transmitting microcurrents into the cells, no current flows when wearing socks. However, this product measures the contact resistance of the feet through its own Hydrating Conductive System(HCS), calculates the presence and thickness of socks, and supplies as much moisture as necessary for electrical conduction. By applying the peristaltic pump of the hemodialysis device, fine supply is possible and measurement is possible without harming the user experience.

The InBody AI scale helps people of various body types around the world, including specialized body types, maintain a natural and comfortable posture with an automatically adjusted display that follows the user's gaze and an ergonomically designed electrode handle and footrest. The sturdy and wide footrest and the long stroke in the height meter allow people from children as tall as 4.2ft to people as tall as 6.6ft to easily climb onto the device and measure without any burden. The upper body of the device, which is equipped with a low-noise BLDC motor and linear rail, minimizes the movement noise of the display to the extreme, maintaining a high user experience. The clean and simple UI design that does not require any input eases the user’s concerns and tension regarding the medical data analysis process of body composition measurement. It is mainly composed of 3D images and videos, which leads to a high level of user understanding and stable posture for using the device, and it is easy for people with language difficulties or disabilities to use it.

a. Digital Health:

Body composition is numerically represented by one's health status so that anyone can easily understand it and is the basis and measure of customized health care. However, there have been the user drop-off of the elderly and child from the information input procedure. In addition, the user drop-off of the impaired who have difficulty maintaining one position for a long time or users who feel the hassle of taking off socks have also occupied a large proportion. The InBody AI scale is a next-generation product that innovatively solves these problems. The measurement begins by simply going up without any input, and it can be measured in only 30 seconds without taking off socks. It provides provides a high level of convenience and accessibility as if standing on a scale, and effectively prevents physical and psychological defectors. These factors provide an opportunity for the technology underprivileged who are faced with limitations in the measurement procedure as above to benefit from digital health technology. The high accessibility of this product will break down the perception that the field is for smart young people and lead a culture that brings digital health technology closer to everyone’s daily lives.

b. Fitness:

Existing body composition analyzers have experienced many user drop-off due to the inconvenience of long measurement times, information input, and taking off socks. Although products are widely distributed to various facilities, many users simply measure their weight and leave, and this has led to a lack of awareness of the necessity of body composition analysis, despite the fact that it is essential in the fitness field. The InBody AI scale is a next-generation product that combines height measurement functions, specialized electrode systems, and AI with existing devices, and has innovatively solved these problems. Measurement begins simply by stepping up and holding the handle without any input, and can be easily measured in just 30 seconds without taking off socks. This provides high convenience and accessibility for men and women of all ages as if they were standing on a scale, and effectively prevents physical and psychological dropouts along with a dramatically reduced measurement time. These factors will greatly increase people's interest in and accessibility to body composition analysis and, furthermore, self-health management, and will contribute to creating a more correct and active fitness culture.

c. Home Appliance:

Although adolescent growth and health management are very important tasks for both parents and children, effectively monitoring them is often difficult. However, InBody AI scale innovatively solves this problem through a groundbreaking change to the existing body composition measurement method. The most notable change is that children can easily measure their body composition without the help of their parents. It completely eliminates the existing cumbersome manual input, encouraging children to develop the habit of monitoring their own health status. In addition to the short measurement time of 30 seconds, it induces familiar recognition and daily use and frequency, just like a scale. This high accessibility helps children accept body composition as a part of their daily life and fosters an attitude of managing their own body composition from an early age. This allows them to recognize the importance of active health management early on, and parents can efficiently manage their children's height and health information. It encourages parenting that follows the growth and changes of children, while enabling wise and systematic parenting through AI-tailored services. InBody AI scale will go beyond simply recording children's growth and become an essential tool for continuous health management.