基于改进YOLO v5的苹果采摘机器人目标检测方法

胡仕林; 陈伟; 张境锋; 魏庆宇; 金学广

doi:10.13427/j.cnki.njyi.2024.06.009

基于改进YOLO v5的苹果采摘机器人目标检测方法

1. 江苏科技大学电子信息学院;
2. 常州信息职业技术学院

基金项目:

江苏省现代农业项目(BE2020406)

常州市科技支撑计划(农业)项目(CE20212025)

镇江市国际科技合作项目(GJ202009)

详细信息

作者简介:
胡仕林(1997-),男,安徽淮南人,硕士研究生,(E-mail)h1072226109@126.com

通讯作者:
陈伟(1981-),男,江苏镇江人,副教授,硕士生导师,(E-mail)cwchenwei@aliyun.com

中图分类号: S225;TP242;TP391.41
计量
- 文章访问数: 8
- HTML全文浏览量: 0
- PDF下载量: 0
出版历程
- 收稿日期: 2022-06-19
- 刊出日期: 2024-05-31

Target Detection Method of Apple Harvesting Robot Based on Improved YOLO v5

1. School of Electronic and Information, Jiangsu University of Science and Technology;
2. Changzhou Information Vocational and Technical College

摘要

摘要: 为实现复杂环境中苹果采摘机器人目标快速检测，克服传统YOLO v5网络结构复杂、计算性能弱的缺点，提出了一种基于深度可分离卷积YOLO v5的采摘机器人目标检测方法。在采集苹果样本图像并制作实验数据集后，进行模型训练和测试，引入深度可分离卷积YOLO v5网络对苹果图像进行特征提取，解决了网络中参数冗余问题，提高了采摘机器人的识别速度；采用CIoU-Loss损失函数和DIoU-NMS非极大值抑制方法，对损失函数进行优化，提升了机器人视觉系统对苹果的定位精度。机器人采摘试验结果表明：算法检测精度达95.8%,检测速度达53帧/s,机器人单次采摘时间为4.7s,采摘成功率达93.9%。检测方法在减少模型参量的同时可保证检测精度和效率，具有较强的工程实用性。
- 苹果采摘机器人 /
- 目标检测 /
- YOLOv5 /
- 深度可分离卷积
Abstract: In order to achieve rapid detection of apple picking robot targets in complex environments, and overcome the shortcomings of traditional YOLO v5 network structure and weak computing performance, a target detection method based on depthwise separable convolution YOLO v5 is proposed. After collecting apple sample images and making experimental data sets, model training and testing are performed, and a depthwise separable convolutional YOLO v5 network is introduced to extract features from apple images, which solves the problem of parameter redundancy in the network and improves the recognition speed of picking robots; The CIoU-Loss loss function and the DIoU-NMS non-maximum suppression method are used to optimize the loss function and improve the positioning accuracy of the robot vision system for apples. The results of the robot picking experiment show that the detection accuracy of the algorithm is 95.8%, the detection speed is 53 frames per second, the single picking time of the robot is 4.7s, and the picking success rate is 93.9%. The detection method can ensure detection accuracy and efficiency while reducing model parameters, and has strong engineering practicability.
- apple picking robot /
- object detection /
- YOLO v5 /
- depthwise separable convolution

HTML全文

参考文献(18)

[1]	KELMAN E E,LINKER R.Vision-based localisation of mature apples in tree images using convexity[J].Biosystems engineering,2018,11(8):174-185.
[2]	KAPACH K,BARNEA E,MAIRON R,et al.Computer vision for fruit harvesting robots-state of the art and challenges ahead[J].International journal of computational vision and robotics,2018,3(2):4-34.
[3]	刘晓洋,赵德安,贾伟宽,等.基于超像素特征的苹果采摘机器人果实分割方法[J].农业机械学报,2019,50(3):15-23.
[4]	吕继东,赵德安.苹果采摘机器人目标果实快速跟踪识别方法[J].农业机械学报,2017,45(1):65-72.
[5]	FAN P,LANG G D,GUO P J,et al.Multi-feature patch -vased segmentation technique in the gray-centered RGB color space for improved apple target recognition[J].Journal of technology and science,2020,14(11):214-223.
[6]	YING B Y,XU Y C,ZHANG S,et al.Weed detection in images of carrot fields based on improved YOLO v4[J].Traitement du signal,2021,38(2):384-418.
[7]	RAKUN J,STAJNKO D,ZAZULA D.Detecting fruits in natural scenes by using spatial-frequency based texture analysis and multiview geometry[J].Computers and electronics in agriculture,2022,76(1):80-88.
[8]	AGELOPOLOU A D,BOCHTIS D,FOUNTAS,et al.Yield prediction in apple orchards based on image processing[J].Precision agriculture,2021,12(3):448-456.
[9]	D Wu,S Lv,M Jiang.Using channel pruning-based YOLO v4 deep learning algorithm for the real-time and accurate detection of apple flowers in natural environments[J].Computers and electronics in agriculture,2021,24(6):737-739.
[10]	WEI JI,XIANGLI MENG,YUN TAO,et al.Fast segmentation of colour apple image under all-weather conditions for vision recognition of picking robots[J].International journal of advanced robotic systems,2019,13(3):86-87.
[11]	E K NYARKO,I VIDOVIC,K Cupec.A nearest neighbor approach for fruit recognition in RGB-D images based on detection of convex surfaces[C]//Proceedings of the AAAI Conference on Artificial Intelligence.Hawaii:AAAI Press,2020:893-907.
[12]	MEHMET TURAN,YASIN ALMALIOGLU,HELDER ARAUJO,et al.Deep EndoVO:a recurrent convolutional neural network based visual odometry approach for endoscopic capsule robots[J].Neurocomputing,2019,15(4):273-281.
[13]	GIRSHICK R,DONAHUE J,DARRELL T,et al.Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the 2019 IEEE Conference on Computer Vision and Pattern Recognition.Columbus:IEEE,2019:580-587.
[14]	GAO F F,FU L S,ZHANG X,et al.Multi-class fruit plant detection for apple in SNAP system using faster R-CNN[J].Computers and electronics in agriculuture,2021,17(6):152-161.
[15]	S.KHAN,H.RAHMANI,AND S.A.shah.A guide to convolutional neural networks for computer vision[J].Synthesis lectures on computer vision,2022,17(8):112-124.
[16]	LIU W,ANGUELOV D,ERHAN D,et al.Ssd:Single shot multibox detector[C]//European Conference on Computer Vision.Amsterdam:Springer,Cham,2021:21-37.
[17]	REDMON J,DIVVALA S,GIRSHICK R,et al.You only look once:unified,real-time object detection[C]//Proceedings of the 2019 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.Las Vegas:IEEE,2019:779-788.
[18]	L FU.Fast and accurate detection of kiwifruit in orchard using improved YOLOv4-tiny model[J].Precision agriculture,2020,38(3):68-69.