VoLearn

Xia, Chengshuo; Fang, Xinrui; Arakawa, Riku; Sugiura, Yuta

doi:10.1145/3534576

Cited by 10 publications

References 60 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

ConvNeXtPose: A Fast Accurate Method for 3D Human Pose Estimation and Its AR Fitness Application in Mobile Devices

Nguyen,

Kim,

et al. 2023

IEEE Access

View full text Add to dashboard Cite

In general, 3D human-pose estimation requires high-performance computing resources. Existing methods working on mobile devices trade off accuracy in return for increased efficiency, often making the estimation accuracy far from sufficient for developing serious applications. In this paper, we present a mobile 3D humanpose estimation model, achieving real-time performances with a well-designed balance between efficiency and accuracy. As the backbone, our model leverages the cutting-edge ConvNeXt architecture, renowned for its feature extraction capabilities. We enhance its performance through strategic architectural modifications and incorporation of depthwise separable convolutions in the upsampling module. The experiments made with the Human3.6M dataset show that the accuracy delivered by our model is comparable to that of the stateof-the-art models, consuming significantly fewer computational resources. To showcase the practicality of our model, we present a prototype of an AR fitness application. Built upon our 3D human pose estimation model, it helps trainees recreate trainers' poses from reference images. The effectiveness of the application is validated via experiments and evaluations. The source code can be found at: https://github.com/medialab-ku/ ConvNeXtPose.

show abstract

ConvNeXtPose: A Fast Accurate Method for 3D Human Pose Estimation and Its AR Fitness Application in Mobile Devices

Nguyen,

Kim,

et al. 2023

IEEE Access

View full text Add to dashboard Cite

show abstract

Virtual Sensors With 3D Digital Human Motion for Interactive Simulation

Xia,

Sugiura

2023

Computer

Self Cite

View full text Add to dashboard Cite

This article presents a design of "virtual sensors" to collect low-dimensional sensor data from 3D digital human motion and create real-world applications through the "interactive simulation." It provides an opportunity to decrease dependency on real-world data requirements and gives more flexibility in the corresponding human activity-related applications. Using sensors and machine learning (ML) to recognize human motion and build the activity-related applications has been successfully applied to assist people's daily lives, such as gesture interaction, motion tracking, and exergames. 1,2 As in Zhu et al., 2 multimodal sensors have been used in various types of activity recognition systems. However, such systems have shown bottlenecks; that is, real sensor datasets are always required as prerequisites to train classifiers. As a result, the developed applications are mainly limited by the dataset's characteristics, including data modalities, sensor characteristics, and data types.

show abstract