Scholarlink

Yoga serves as an excellent means of physical exercise and holds promise as an application in personal healthcare. Achieving proper body postural alignment is a crucial aspect during the practice of yogic asanas. An incorrect asana can cause joint and ligament strains, and even backbone problem. Thus, it is essential to supervise the proper performance of yoga postures during different asanas. Recent progress in computer vision algorithms and sensor technology has made it feasible to classify yoga postures and analyse movements automatically. In order to classify the yoga pose, four deep neural network models are used in this paper, they are VGG-16, VGG-19, ResNet 50 and ResNet101, out of which ResNet101 architecture outperformed and yielded excellent accuracy of 96.84%.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Efficient Yoga Pose Classification Using Deep Neural Networks

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • K. Divya,
  • T. Sree Sharmila

摘要

Yoga serves as an excellent means of physical exercise and holds promise as an application in personal healthcare. Achieving proper body postural alignment is a crucial aspect during the practice of yogic asanas. An incorrect asana can cause joint and ligament strains, and even backbone problem. Thus, it is essential to supervise the proper performance of yoga postures during different asanas. Recent progress in computer vision algorithms and sensor technology has made it feasible to classify yoga postures and analyse movements automatically. In order to classify the yoga pose, four deep neural network models are used in this paper, they are VGG-16, VGG-19, ResNet 50 and ResNet101, out of which ResNet101 architecture outperformed and yielded excellent accuracy of 96.84%.