PixE: Home Fitness Method Using Machine Learning with Smartphone

Abstract

The COVID-19 outbreak has caused worldwide confusion. However, in the Republic of Korea, it has overcome the situation by using a variety of communication infrastructures. Especially in Korea, a new word Homt appeared. Home fitness is referred to as ‘home training' in Korea and is called Homt for short. It is a word for a new method of home fitness which has emerged based on Internet infrastructure. This is a service made possible on a fast network in Korea, where a great deal of data is transmitted at high speed. Our research goal is to create fitness game, PixE, using a smartphone camera that everybody has. Basically, it is a game of strength exercises such as squats and lunges. It uses machine learning technology to exchange information over a fast network in Korea while minimizing the burden of poor smartphone resources. It is based on the detection of human movements by getting images from a smartphone in real time on the server. Motion recognition and a variety of effects are created using extracted images. The final image is pixelated and sent back to the smartphone. With the edited image by using ML, the user himself becomes a main character in the fitness game. It is a fitness game for Homt that can be easily used anytime, anywhere. If the user only has a smartphone, that's all! In particular, it has a user focus and be immersed in it by becoming the main character that appears directly into the game. Before the official release, we demonstrated with two groups, exercise experts and laymen. We found that the effects of the exercise were no different than the existing effects. We hope to create an environment where you can easily home fitness anytime, anywhere, not in a special situation such as COVID-19.