Shizuoka University
Okabe Lab

Practice Support System for Stretched Backward Somersaults using Object Detection and Posture Estimation

Kota Takahashi

Shizuoka University

Makoto Okabe

Shizuoka University

研究のTeaser画像

A sequence of photographs visualizing the flow of a artistic gymnastics performance. It can be created quickly and automatically by using object detection technology to detect human parts.

Abstract

We propose a system to support the practice of streched backward somersaults in floor exercise of artistic gymnastics using object detection and posture estimation. The system detects the part of the performance videos where the person is and automatically creates a sequence of photos by cropping the area. The system also aims to analyze the posture and movement of the performance using the coordinates of joint points obtained through posture estimation and to suggest areas that need to be corrected, providing an environment for efficient technical improvement.

Paper

Interim Abstract

Video

BibTeX citation

    @Misc{takahashi.kota.20_M1,
        author       = {Kota Takahashi},
        title        = {Practice Support System for Stretched Backward Somersaults using Object Detection and Posture Estimation},
        howpublished = {Master's First-Year Research Presentation},
        month        = 1,
        year         = 2025,
    }