基于深度学习的花椒采摘器视觉识别
张志会1,张勇2,刘雪垠1,郭恒3,杨永林3
(1.四川省机械研究设计院(集团)有限公司,四川 成都 610063;2.四川航天烽火伺服控制技术有限公司,四川 成都 610199;3.中国机械工业第一建设有限公司,四川 德阳 618000)
关键词:对于智能花椒采摘器中机器视觉部分在花椒枝干识别与采摘定位上的不足,本文通过将深度学习技术中的卷积神经网络模型与注意力机制这两种模型运用到智能花椒采摘器的机器视觉部分以提高采摘器的识别功效。结果显示,经过优化后的卷积神经网络算法训练使采摘器对花椒簇的整体识别准确率由52.3%提高至96.7%,同时通过注意力机制算法提升了机器视觉对花椒树主枝干识别的抗干扰能力,帮助采摘器更加准确的判断出采摘点的位置。通过以上两种模型验证了深度学习技术在提高花椒采摘器机器视觉的算法准确性与抗干扰能力的有效性。
摘要:花椒采摘器;机器视觉;卷积神经网络模型;注意力机制
中图分类号:S22 文献标志码:A doi:10.3969/j.issn.1006-0316.2021.11.003
文章编号:1006-0316 (2021) 11-0017-08
Visual Recognition of Pepper Picker Based on Deep Learning
ZHANG Zhihui1,ZHANG Yong2,LIU Xueyin1,GUO Heng3,YANG Yonglin3
( 1.Sichuan Machinery Research and Design Institute (Group) Co., Ltd., Chengdu 610063, China;2.Sichuan Aerospace Fenghuo Servo Control Technology Co., Ltd., Chengdu 610199, China;3.The First Construction of China Mechanical Industry Co., Ltd., Deyang 618000, China )
Abstract:There are deficiencies in machine vision of intelligent pepper picker in the pepper branch recognition and picking positioning. This paper applies convolution neural network model and attention mechanism of deep learning technology to the machine vision part of the intelligent pepper picker, so as to improve the recognition effect of the picker. The results show that after the optimized convolution neural network algorithm training, the overall recognition accuracy of pepper cluster is improved from 52.3% to 96.7%. At the same time, the attention mechanism algorithm improves the anti-interference ability of machine vision for pepper main branch recognition, which helps the picker to judge the picking point more accurately. The above two models verify the effectiveness of deep learning technology in improving the algorithm accuracy and anti-interference ability of pepper picker machine vision.
Keywords:pepper picker;machine vision;convolutional neural network model;attention mechanism
———————————————
收稿日期:2021-03-18
基金项目:四川省科技计划项目——簇状果实采摘机器人关键技术研究及研制(2021YFN0020)
作者简介:张志会(1982-),男,满族,内蒙古赤峰人,硕士,工程师,主要研究方向为机械电子工程,E-mail:99249829@qq.com。
 

 

设为首页  |  加入收藏    |   免责条款
《机械》杂志版权所有     Copyright©2008-2012 Jixiezazhi.com All Rights Reserved 

  电话:028-85925070    传真:028-85925073    E-mail:jixie@vip.163.com

地址:四川省成都锦江工业开发区墨香路48号   邮编:610063

蜀ICP备08103512号

Powered by PageAdmin CMS