The tea buds detection and yield estimation method based on optimized YOLOv8

文献类型： 外文期刊

第一作者： Zhou, Changan

作者： Zhou, Changan;Zhu, Youxu;Zhang, Jialin;Zhang, Kaixing;Ding, Zhaotang;Jiang, Wenjuan

作者机构：

关键词： Tea yield estimation; Object detection; Computer vision; YOLOv8

期刊名称：SCIENTIA HORTICULTURAE （ 影响因子：4.2； 五年影响因子：4.6 ）

ISSN： 0304-4238

年卷期： 2024 年 338 卷

页码：

收录情况： SCI

摘要： The existing methodology for detecting and estimating tea bud yields heavily relies on cultivation experience, resulting in a delay in obtaining accurate information regarding tea bud yields. This delay poses challenges in providing timely data support for tea leaf growth management during the early stages of tea production. To address this issue, we propose a novel approach for tea bud detection and yield estimation. Initially, image data is gathered from the tea plantation area to train an optimized YOLOv8 model, enabling precise detection and enumeration of tea buds within the specified region. Ablation experiment is conducted to validate the effectiveness of the model improvements. Subsequently, data associating tea bud quantities with yields are collected, and a yield estimation mathematical model is derived through the partial least squares (PLS) method. Finally, the target detection model is integrated with the yield estimation model. By leveraging data such as the tea bud count within the target area, estimated area, and total planting area, the system accomplishes the estimation of tea bud yields. The outcomes of research are as follows: 1) The optimized YOLOv8 model achieves a recognition accuracy of 95.9 % for tea buds, boasting a recall rate of 85.5 % and an average precision of 91.7 %. 2) The relative error between the estimated yield and actual yield, as determined by the proposed method in this study, stands at 10.28 %. The tea bud detection and yield estimation methodology introduced in this research offer a convenient and expeditious means to estimate tea bud yields. It provides valuable data support to tea growers, facilitating pertinent production planning within the tea industry.

分类号：

  • 相关文献
作者其他论文 更多>>

微信小程序