基于YOLOV3优化模型的复杂场景下茶树嫩芽识别

doi:10.3969/j.issn.1004-1524.2021.09.18

浙江农业学报 ›› 2021, Vol. 33 ›› Issue (9): 1740-1747.DOI: 10.3969/j.issn.1004-1524.2021.09.18

基于YOLOV3优化模型的复杂场景下茶树嫩芽识别

张晴晴¹(), 刘连忠¹^,^*(), 宁井铭²^,³, 吴国栋¹, 江朝晖¹, 李孟杰¹, 李栋梁¹

1.安徽农业大学信息与计算机学院,安徽合肥 230036
2.安徽农业大学茶与食品科技学院,安徽合肥 230036
3.茶树生物学与资源利用国家重点实验室,安徽合肥 230036

收稿日期:2020-07-18 出版日期:2021-09-25 发布日期:2021-10-09
通讯作者: 刘连忠
作者简介:* 刘连忠,E-mail: lzliu@ahau.edu.cn
张晴晴(1995—),女,安徽亳州人,硕士研究生,主要研究方向为图像处理、图像识别。E-mail: qingqz_ah@163.com
基金资助:
茶树生物学与资源利用国家重点实验室开放基金(SKLTOF20160202);安徽省高校自然科学研究项目重点项目(KJ2017A151);安徽省高校自然科学研究重大项目(KJ2019ZD20);国家重点研发计划(2016YFD0200900)

Tea buds recognition under complex scenes based on optimized YOLOV3 model

ZHANG Qingqing¹(), LIU Lianzhong¹^,^*(), NING Jingming²^,³, WU Guodong¹, JIANG Zhaohui¹, LI Mengjie¹, LI Dongliang¹

1. College of Information and Computer Science, Anhui Agricultural University, Hefei 230036, China
2. College of Tea and Food Science and Technology, Anhui Agricultural University, Hefei 230036, China
3. State Key Laboratory of Tea Plant Biology and Resource Utilization, Hefei 230036, China

Received:2020-07-18 Online:2021-09-25 Published:2021-10-09
Contact: LIU Lianzhong

摘要/Abstract

摘要：

茶叶智能采摘的关键技术之一是待采摘嫩芽的识别,而嫩芽大小、环境光照、拍摄角度等因素都会给嫩芽的精准识别带来困难。针对复杂场景下传统茶树嫩芽识别方法准确率低的问题,文章提出一种基于YOLOV3深度卷积模型的识别方法,并通过增加SPP模块优化模型,提高模型对茶树嫩芽的识别能力。实验结果表明,YOLOV3模型和YOLOV3优化模型均能在复杂场景下实现茶树嫩芽识别,且YOLOV3优化模型的平均精度均值mAP比YOLOV3模型提高3.5百分点,达到91%,说明YOLOV3优化模型能够更好地应用于自然场景下的茶树嫩芽识别。

关键词: 深度学习, 卷积神经网络, 茶树嫩芽识别

Abstract:

Tea buds recognition is one of the key technologies for intelligent tea buds picking, and the size of buds, environmental lighting, imaging angle and other factors will bring difficulties for precise buds identification. In order to solve the problem of low accuracy of traditional tea buds recognition in complex scenes, a tea buds recognition method based on YOLOV3 deep convolution model was proposed. By adding a SPP module into the YOLOV3 model, an optimized YOLOV3 model was designed to further improve the recognition ability of tea buds. The results showed that YOLOV3 and optimized YOLOV3 models could both realize tea buds recognition in complex scenes, and the average accuracy mean (mAP) of the optimized YOLOV3 model reached 91%, which was 3.5 percentage points higher than YOLOV3 model, indicating the optimized YOLOV3 could be well applied to the tea buds identification in natural environments.

Key words: deep learning, convolutional neural network, tea buds recognition

中图分类号:

S571.1
S24

张晴晴, 刘连忠, 宁井铭, 吴国栋, 江朝晖, 李孟杰, 李栋梁. 基于YOLOV3优化模型的复杂场景下茶树嫩芽识别[J]. 浙江农业学报, 2021, 33(9): 1740-1747.

ZHANG Qingqing, LIU Lianzhong, NING Jingming, WU Guodong, JIANG Zhaohui, LI Mengjie, LI Dongliang. Tea buds recognition under complex scenes based on optimized YOLOV3 model[J]. Acta Agriculturae Zhejiangensis, 2021, 33(9): 1740-1747.

图/表 9

图1 茶树图像的采集、裁剪与标记 a,原始茶树图像;b,裁剪后的茶树图像;c,茶树嫩芽标记。

Fig.1 Collection, cropping and marking of tea tree images a,Original tea tree image;b,Cropped tea tree image;c,Tea bud marking.

图2 实验整体流程图

Fig.2 Overall flow chart of experiment

图3 图像预处理 a.预处理前; b.预处理后。

Fig.3 Image pre-processing a.Before pretreatment;b.After pretreatment.

图4 YOLOV3优化模型的网络结构

Fig.4 The network structure of optimized YOLOV3 model

图5 两种模型的mAP值随训练次数的变化曲线

Fig.5 The mAP values of the two models change curve with the training times

表1 两种模型的性能对比

Table 1 Performance comparison of the two models

模型种类 Model type	平均精度均值mAP Mean average precision/%	召回率 The recall rate/%	平均检测时间 Average detection time/s
YOLOV3模型YOLOV3 model	87.5	71	0.375 5
YOLOV3优化模型Optimized YOLOV3 model	91.0	75	0.387 2

图6 两种模型在不同嫩芽数量的识别结果

Fig.6 The recognition results of the two models with different number of buds

图7 两种模型在不同光照条件下的识别结果

Fig.7 The recognition results of the two models under different lighting conditions

图8 两种模型在不同拍摄角度下的识别结果

Fig.8 Recognition results of the two models from different shooting angles

参考文献 30

[1]	寇楠楠. 茶叶对神经系统作用及健康伦理视域下茶之健康功能的实现[J]. 茶叶通讯, 2020, 47(1):152-155.
	KOU N N. The effect of tea on the nervous system and the realization of the health function of tea from the perspective of health ethics[J]. Tea Communication, 2020, 47(1):152-155.(in Chinese with English abstract)
[2]	沈童菲, 王森培, 张慧, 等. 中国茶产业国际竞争力评价与发展对策[J]. 四川农业科技, 2019(9):5-11.
	SHEN T F, WANG S P, ZHANG H, et al. Evaluation and development strategy of international competitiveness of Chinese tea industry[J]. Sichuan Agricultural Science and Technology, 2019(9):5-11.(in Chinese)
[3]	刘秋凤, 黄婷婷, 韦瑛璐, 等. 轻便型单人采茶机的采摘效果比较试验[J]. 南方农业, 2020, 14(16):65-67.
	LIU Q F, HUANG T T, WEI Y L, et al. Comparison experiment of picking effect of portable single tea-picking machine[J]. South China Agriculture, 2020, 14(16):65-67.(in Chinese)
[4]	黄丹娟, 王红娟, 陈勋, 等. 茶园机采树冠培育制度研究进展[J]. 茶叶通讯, 2020, 47(2):192-197.
	HUANG D J, WANG H J, CHEN X, et al. Research progress on cultivation system of machine harvested crown in tea gardens[J]. Journal of Tea Communication, 2020, 47(2):192-197.(in Chinese with English abstract)
[5]	任冬法. 浅谈机械自动化技术在茶叶生产加工方面的应用[J]. 福建茶叶, 2020, 42(6):29-30.
	REN D F. Application of mechanical automation technology in tea production and processing[J]. Tea in Fujian, 2020, 42(6):29-30.(in Chinese)
[6]	夏华鹍, 史必高, 黄海霞, 等. 图像处理在茶叶嫩芽智能采摘中的应用进展[J]. 安徽农学通报, 2019, 25(9):133-134.
	XIA H K, SHI B G, HUANG H X, et al. Application progress of image processing in intelligent picking of tea sprouts[J]. Anhui Agricultural Science Bulletin, 2019, 25(9):133-134.(in Chinese with English abstract)
[7]	唐仙, 吴雪梅, 张富贵, 等. 基于阈值分割法的茶叶嫩芽识别研究[J]. 农业装备技术, 2013, 39(6):10-14.
	TANG X, WU X M, ZHANG F G, et al. Contrastive research on tender tea recognition based on multiple threshold segmentation methods[J]. Agricultural Equipment & Technology, 2013, 39(6):10-14.(in Chinese with English abstract)
[8]	汪建. 结合颜色和区域生长的茶叶图像分割算法研究[J]. 茶叶科学, 2011, 31(1):72-77.
	WANG J. Segmentation algorithm of tea combined with the color and region growing[J]. Journal of Tea Science, 2011, 31(1):72-77.(in Chinese with English abstract)
[9]	GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]// 2014 IEEE Conference on Computer Vision and Pattern Recognition. Columbus, OH, USA. IEEE, 2014: 580-587.
[10]	LECUN Y, BENGIO Y, HINTON G. Deep learning[J]. Nature, 2015, 521(7553):436-444. DOI URL
[11]	郭丽丽, 丁世飞. 深度学习研究进展[J]. 计算机科学, 2015, 42(5):28-33.
	GUO L L, DING S F. Research progress on deep learning[J]. Computer Science, 2015, 42(5):28-33.(in Chinese with English abstract)
[12]	GIRSHICK R. Fast R-CNN[C]// 2015 IEEE International Conference on Computer Vision (ICCV). Santiago, Chile. IEEE, 2015: 1440-1448.
[13]	REN S Q, HE K M, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6):1137-1149. DOI URL
[14]	房靖晶, 成金勇. 基于改进的Faster R-CNN的目标检测与识别[J]. 图像与信号处理, 2019, 8(2):43-50.
	FANG J J, CHENG J Y. Target detection and recognition based on improved Faster R-CNN[J]. Image and Signal Processing, 2019, 8(2):43-50.(in Chinese with English abstract)
[15]	GU J X, WANG Z H, KUEN J, et al. Recent advances in convolutional neural networks[J]. Pattern Recognition, 2018, 77:354-377. DOI URL
[16]	SHINDE S, KOTHARI A, GUPTA V. YOLO based human action recognition and localization[J]. Procedia Computer Science, 2018, 133:831-838. DOI URL
[17]	LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot MultiBox detector[M]//Computer vision: ECCV 2016. Cham: Springer International Publishing, 2016: 21-37.
[18]	林君宇, 李奕萱, 郑聪尉, 等. 应用卷积神经网络识别花卉及其病症[J]. 小型微型计算机系统, 2019, 40(6):1330-1335.
	LIN J Y, LI Y X, ZHENG C W, et al. Classifying flowers and their diseases by using convolutional neural network[J]. Journal of Chinese Computer Systems, 2019, 40(6):1330-1335.(in Chinese with English abstract)
[19]	刘小刚, 范诚, 李加念, 等. 基于卷积神经网络的草莓识别方法[J]. 农业机械学报, 2020, 51(2):237-244.
	LIU X G, FAN C, LI J N, et al. Identification method of strawberry based on convolutional neural network[J]. Transactions of the Chinese Society for Agricultural Machinery, 2020, 51(2):237-244.(in Chinese with English abstract)
[20]	YAN L, PANG L, WANG H, et al. Recognition of different Longjing fresh tea varieties using hyperspectral imaging technology and chemometrics[J]. Journal of Food Process Engineering, 2020, 43(4):e13378.
[21]	吕军, 夏华鹍, 方梦瑞, 等. 基于AlexNet的茶叶嫩芽状态智能识别研究[J]. 黑龙江八一农垦大学学报, 2019, 31(2):72-78.
	LYU J, XIA H K, FANG M R, et al. Research on intelligent identification of tea sprouts state based on AlexNet[J]. Journal of Heilongjiang Bayi Agricultural University, 2019, 31(2):72-78.(in Chinese with English abstract)
[22]	孙肖肖, 牟少敏, 许永玉, 等. 基于深度学习的复杂背景下茶叶嫩芽检测算法[J]. 河北大学学报(自然科学版), 2019, 39(2):211-216.
	SUN X X, MU S M, XU Y Y, et al. Detection algorithm of tea tender buds under complex background based on deep learning[J]. Journal of Hebei University (Natural Science Edition), 2019, 39(2):211-216.(in Chinese with English abstract)
[23]	ZHANG X L, DONG X P, WEI Q J, et al. Real-time object detection algorithm based on improved YOLOv3[J]. Journal of Electronic Imaging, 2019, 28(5):1.
[24]	刘连忠, 李孟杰, 宁井铭. 基于时序巡航图像的茶树生长监测研究[J]. 浙江农业学报, 2020, 32(5):886-896.
	LIU L Z, LI M J, NING J M. Tea plant growth monitoring based on time series cruise images[J]. Acta Agriculturae Zhejiangensis, 2020, 32(5):886-896.(in Chinese with English abstract)
[25]	阮秋琦. 数字图像处理学[M]. 2版. 北京: 电子工业出版社, 2007.
[26]	胡琼, 汪荣贵, 胡韦伟, 等. 基于直方图分割的彩色图像增强算法[J]. 中国图象图形学报, 2009, 14(9):1776-1781.
	HU Q, WANG R G, HU W W, et al. Color image enhancement based on histogram segmentation[J]. Journal of Image and Graphics, 2009, 14(9):1776-1781.(in Chinese with English abstract)
[27]	LIN T Y, DOLLÁR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu, HI, USA. IEEE, 2017: 936-944.
[28]	WANG X L, WANG S, CAO J Q, et al. Data-driven based tiny-YOLOv3 method for front vehicle detection inducing SPP-net[J]. IEEE Access, 2020, 8:110227-110236. DOI URL
[29]	赵德安, 吴任迪, 刘晓洋, 等. 基于YOLO深度卷积神经网络的复杂背景下机器人采摘苹果定位[J]. 农业工程学报, 2019, 35(3):164-173.
	ZHAO D A, WU R D, LIU X Y, et al. Apple positioning based on YOLO deep convolutional neural network for picking robot in complex background[J]. Transactions of the Chinese Society of Agricultural Engineering, 2019, 35(3):164-173.(in Chinese with English abstract)
[30]	周云成, 许童羽, 郑伟, 等. 基于深度卷积神经网络的番茄主要器官分类识别方法[J]. 农业工程学报, 2017, 33(15):219-226.
	ZHOU Y C, XU T Y, ZHENG W, et al. Classification and recognition approaches of tomato main organs based on DCNN[J]. Transactions of the Chinese Society of Agricultural Engineering, 2017, 33(15):219-226.(in Chinese with English abstract)

编辑推荐

Metrics

阅读次数

全文

1779

HTML			PDF

最新录用	在线预览	正式出版	最新录用	在线预览	正式出版
0	0	24	0	0	1755

来源	本网站	其他网站

次数	1401	378
比例	79%	21%

摘要

1135

最新录用	在线预览	正式出版

0	0	1135

来源	本网站	其他网站

次数	1132	3
比例	100%	0%

基于YOLOV3优化模型的复杂场景下茶树嫩芽识别

Tea buds recognition under complex scenes based on optimized YOLOV3 model

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 9

参考文献 30

相关文章 6

编辑推荐

Metrics

本文评价

[1]	鲍烈, 王曼韬, 刘江川, 文波, 明月. 基于卷积神经网络的小麦产量预估方法[J]. 浙江农业学报, 2020, 32(12): 2244-2252.
[2]	蒋丰千, 李旸, 余大为, 孙敏, 张恩宝. 基于Caffe卷积神经网络的大豆病害检测系统[J]. 浙江农业学报, 2019, 31(7): 1177-1183.
[3]	王彦翔, 张艳, 杨成娅, 孟庆龙, 尚静. 基于深度学习的农作物病害图像识别技术进展[J]. 浙江农业学报, 2019, 31(4): 669-676.
[4]	金秀, 卢杰, 傅运之, 王帅, 许高健, 李绍稳. 基于深度卷积神经网络的小麦赤霉病高光谱病症点分类方法[J]. 浙江农业学报, 2019, 31(2): 315-325.
[5]	杨国亮, 许楠, 康乐乐, 龚曼, 洪志阳. 基于参数指数非线性残差神经网络的脐橙病变叶片识别[J]. 浙江农业学报, 2018, 30(6): 1073-1081.
[6]	张善文, 黄文准, 尤著宏. 基于物联网和深度卷积神经网络的冬枣病害识别方法[J]. 浙江农业学报, 2017, 29(11): 1868-1874.