Scholarlink

<p>Volleyball training robots play an important role in modern sports training, and their automatic serving technology can simulate serving modes in different scenarios, providing athletes with diverse training programs. However, due to the characteristics of volleyball, such as small size, fast speed, and susceptibility to occlusion and noise interference in complex backgrounds, traditional object detection methods are hard to meet their real-time and accuracy requirements. An automatic serving method for volleyball training robots is proposed based on You Only Look Once v5 and Hough transform. By introducing convolutional block attention module to optimize feature extraction and focus on key areas, a weighted bi-directional feature pyramid network is taken to fuse multi-scale features, and gradient optimized Hough transform is used to optimize the accuracy of target localization. The model outperformed the comparison model in accuracy, root mean square error, recall, and F1-score, with a detection accuracy of 92.3% and a root mean square error of only 0.18. The success rate in capturing dynamic targets reached 95.6%, and the trajectory tracking error was reduced to 0.1&#xa0;m, significantly improving the automation level in volleyball sports. The method has good accuracy and real-time performance, providing an efficient and accurate automatic serving method for volleyball training robots.</p>

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

Automatic serving method of volleyball training robot based on improved YOLOv5 and improved Hough transform

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年8月1日

查看全文
  • Tao Sun,
  • Xiaolong He,
  • Jiajun Zhang

摘要

Volleyball training robots play an important role in modern sports training, and their automatic serving technology can simulate serving modes in different scenarios, providing athletes with diverse training programs. However, due to the characteristics of volleyball, such as small size, fast speed, and susceptibility to occlusion and noise interference in complex backgrounds, traditional object detection methods are hard to meet their real-time and accuracy requirements. An automatic serving method for volleyball training robots is proposed based on You Only Look Once v5 and Hough transform. By introducing convolutional block attention module to optimize feature extraction and focus on key areas, a weighted bi-directional feature pyramid network is taken to fuse multi-scale features, and gradient optimized Hough transform is used to optimize the accuracy of target localization. The model outperformed the comparison model in accuracy, root mean square error, recall, and F1-score, with a detection accuracy of 92.3% and a root mean square error of only 0.18. The success rate in capturing dynamic targets reached 95.6%, and the trajectory tracking error was reduced to 0.1 m, significantly improving the automation level in volleyball sports. The method has good accuracy and real-time performance, providing an efficient and accurate automatic serving method for volleyball training robots.