复杂茶园场景的轻量化茶芽采摘点检测模型
CSTR:
作者:
作者单位:

(东北林业大学计算机与控制工程学院,哈尔滨 150040)

作者简介:

张澎涛(1980—),副教授,博士研究生,硕士生导师,主要从事计算机视觉、智能检测与故障诊断领域的研究;张佳薇(1975—),教授,博士研究生,博士生导师,主要从事多传感器数据融合、林业信息智能检测、林业物联网等领域的研究。

通讯作者:

中图分类号:

TP391. 41

基金项目:


A Lightweight Detection Model for Tender Tea Shoot Picking Points in Complex Tea Garden Scenes
Author:
Affiliation:

(College of Computer and Control Engineering,Northeast Forestry University,Harbin 150040,CHN)

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    针对光电成像中微小、遮挡、高速目标的实时定位难题,传统姿态估计模型在保持微小关键点特征与轻量化推理之间难以平衡。文章以复杂茶园中的茶芽采摘点定位为典型验证场景,提出一种基于改进 YOLOv8n-Pose的轻量化姿态估计模型。首先,设计并行化补丁感知注意力模块,增强网络对细微连接处特征的捕获能力;其次,构建多源特征融合模块与递进式双向循环颈部架构,实现全尺度定位特征的高效互联与互补增强;最后,引入基于交并比加权的样本动态加权损失,提升模型对遮挡、模糊等困难样本的学习关注度。在自建数据集上的实验结果表明,关键点检测平均精度均值达 79. 1%,参数量仅为 2.59 M,计算量为 8. 3 GFLOPs。经 TensorRT加速后,该模型在边缘设备上实现 106. 84 FPS的推理速度。

    Abstract:

    To address the common challenge of real-time,high-precision localization of tiny,occluded,and fast-moving targets in photoelectric imaging systems,conventional pose estimation models often fail to balance the detection of subtle features with lightweight deployment. This study takes the picking point localization of tender tea shoots in complex tea gardens as a representative scenario,and a lightweight pose estimation model based on an improved YOLOv8n-Pose is proposed. Firstly,a parallelized patch-aware attention module is designed to enhance the capture of delicate features at junctions. Secondly,a multi-source feature fusion module and a progressive bidirectional recurring neck architecture are constructed for efficient full-scale feature interconnection. Finally,a slide-weighted loss function based on IoU is introduced to focus learning on hard samples. Experiments on a custom dataset demonstrated a keypoint mAP of 79. 1%,with only 2. 59 M parameters and 8. 3 GFLOPs. Deployed on an edge device with TensorRT,the model achieved 106. 84 FPS.

    参考文献
    相似文献
    引证文献
引用本文

张澎涛,于帅,黄建平,张宪奇,张佳薇.复杂茶园场景的轻量化茶芽采摘点检测模型[J].半导体光电,2026,47(5):827-836. ZHANG Pengtao, YU Shuai, HUANG Jianping, ZHANG Xianqi, ZHANG Jiawei. A Lightweight Detection Model for Tender Tea Shoot Picking Points in Complex Tea Garden Scenes[J].,2026,47(5):827-836.

复制
分享
相关视频

文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2026-04-26
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2026-10-23
  • 出版日期:
文章二维码