基于YOLACT-RFX模型的穴盘甘蓝苗株分割算法

doi:10.19678/j.issn.1000-3428.0066447

计算机工程 ›› 2023, Vol. 49 ›› Issue (12): 214-223. doi: 10.19678/j.issn.1000-3428.0066447

基于YOLACT-RFX模型的穴盘甘蓝苗株分割算法

王楷¹^,², 韩笑²^,³^,⁴, 朱华吉²^,³^,⁴, 缪祎晟²^,³^,⁴, 吴华瑞¹^,²^,³^,⁴^,*

1. 江苏大学计算机科学与通信工程学院, 江苏镇江 212013
2. 国家农业信息化工程技术研究中心, 北京 100097
3. 北京市农林科学院信息技术研究中心, 北京 100097
4. 农业农村部数字乡村技术重点实验室, 北京 100097

收稿日期:2022-12-05 出版日期:2023-12-15 发布日期:2023-03-22
通讯作者: 吴华瑞
作者简介:
王楷（1997—），男，硕士研究生，主研方向为计算机视觉、农业信息化技术
韩笑，助理研究员、硕士
朱华吉，副研究员、博士
缪祎晟，副研究员、博士
基金资助:
科技创新2030——“新一代人工智能”重大项目(2021ZD0113604); 财政部和农业农村部：国家现代农业产业技术体系资助项目(CARS-23-D07)

Segmentation Algorithm of Plug Cabbage Seedlings Based on YOLACT-RFX Model

Kai WANG¹^,², Xiao HAN²^,³^,⁴, Huaji ZHU²^,³^,⁴, Yisheng MIAO²^,³^,⁴, Huarui WU¹^,²^,³^,⁴^,*

1. School of Computer Science and Communication Engineering, Jiangsu University, Zhenjiang 212013, Jiangsu, China
2. National Engineering Research Center for Information Technology in Agriculture, Beijing 100097, China
3. Information Technology Research Center, Beijing Academy of Agriculture and Forestry Sciences, Beijing 100097, China
4. Key Laboratory of Digital Village Technology, Ministry of Agriculture and Rural Affairs, Beijing 100097, China

Received:2022-12-05 Online:2023-12-15 Published:2023-03-22
Contact: Huarui WU

摘要/Abstract

摘要：

温室作物长势分析是近年来农业信息化领域中的研究热点，目前国内温室多用穴盘育苗的方式，其密集种植的特点和复杂的背景干扰给穴盘苗株的分割识别任务带来挑战。提出一种基于YOLACT-RFX的分割算法实现对穴盘内甘蓝苗株的高精度分割和苗期识别。通过引入递归特征金字塔结构加强甘蓝苗株叶片边缘处的特征提取能力，改进相邻穴盘孔位中相互干扰苗株的分割性能。在递归特征金字塔结构中利用空洞空间金字塔池化结构对尺寸和形状快速变化的甘蓝苗株进行特征识别。最后，融合ResNeXt主干网络提升算法精度，加快模型收敛速度。基于甘蓝苗自建数据集验证所提算法的有效性，实验结果表明，当交并比为0.5时，YOLACT-RFX算法的各类平均精度为84.4%，平均召回率为92.7%，相较于YOLACT算法分别提升了3.6%和3.9%。在同等情况下，分割效果优于MASK-RCNN、SOLO、QueryInst等算法。改进后的YOLACT-RFX算法可实现对不同生长期内甘蓝穴盘苗株的高精度分割，为温室自动化甘蓝苗期管理提供技术基础。

关键词: 分割算法, 甘蓝苗株, 苗期识别, 递归特征金字塔, 空洞空间金字塔池化

Abstract:

In recent years, greenhouse crop growth analysis has become a research focal point in the agricultural informatization field. Currently, the seedling cultivation method with plug trays is predominantly utilized in greenhouses. However, dense planting characteristics and complex background interference introduces challenges for seedling identification and segmentation. To address these challenges, a segmentation algorithm based on YOLACT-RFX is proposed in this study, to effectively achieve high precision segmentation and seedling stage identification of cabbage seedlings in plug trays. The Recursive Feature Pyramid(RFP) structure was introduced to enhance the feature extraction capability at the cabbage seedling leaf edge, and segmentation performance of seedlings with mutual interference in adjacent hole positions was improved. The Atrous Spatial Pyramid Pooling(ASPP) structure was used in the RFP structure to identify cabbage seedling features with rapid changes in size and shape. Subsequently, the ResNeXt backbone network was fused to further improve the algorithm's accuracy and to accelerate model convergence. Based on a field-collected cabbage seedling data set, experimental results demonstrated that when the IoU ratio was 0.5, the average mean precision of the proposed YOLACT-RFX algorithm was 84.4% and the average recall rate was 92.2%. Compared to the original YOLACT algorithm, the YOLACT-RFX method was 3.6 and 3.9% more efficient. Under the same condition, the segmentation results outperformed the MASK-RCNN, SOLO and QueryInst algorithms. Finally, it was confirmed that the proposed YOLACT-RFX algorithm achieved high-precision cabbage seedling segmentation, thus, laying the foundation for automatic cabbage seedling management in greenhouses.

Key words: segmentation algorithm, cabbage seedlings, seedling stage identification, Recursive Feature Pyramid(RFP), Atrous Spatial Pyramid Pooling(ASPP)

王楷, 韩笑, 朱华吉, 缪祎晟, 吴华瑞. 基于YOLACT-RFX模型的穴盘甘蓝苗株分割算法[J]. 计算机工程, 2023, 49(12): 214-223.

Kai WANG, Xiao HAN, Huaji ZHU, Yisheng MIAO, Huarui WU. Segmentation Algorithm of Plug Cabbage Seedlings Based on YOLACT-RFX Model[J]. Computer Engineering, 2023, 49(12): 214-223.

http://www.ecice06.com/CN/Y2023/V49/I12/214

图/表 16

图1 甘蓝苗数据集

Fig.1 Cabbage seedlings data set

图2 标注后图像样本

Fig.2 Annotated image sample

图3 YOLACT-RFX算法结构

Fig.3 Structure of YOLACT-RFX algorithm

图4 FPN与RFP网络结构

Fig.4 Structures of FPN and RFP network

图5 RFP网络的展开结构

Fig.5 Expanded structure of RFP network

图6 ASPP模块的结构

Fig.6 Structure of ASPP module

图7 ResNeXt主干网络的改进

Fig.7 ResNeXt backbone improvements

图8 训练损失曲线

Fig.8 Training loss curve

图9 检测平均精度和分割平均精度训练曲线

Fig.9 Training curves of detection and segmentation mean precision

图10 不同算法的甘蓝苗株分割结果对比

Fig.10 Comparison of segmentation results of cabbage seedlings by different algorithms

参考文献 31

1	孟庆宽, 张漫, 叶剑华, 等. 基于轻量二阶段检测模型的自然环境多类蔬菜幼苗识别. 农业机械学报, 2021, 52(10): 282- 290. doi: 10.6041/j.issn.1000-1298.2021.10.029
	MENG Q K, ZHANG M, YE J H, et al. Identification of multiple vegetable seedlings based on two-stage lightweight detection model. Transactions of the Chinese Society for Agricultural Machinery, 2021, 52(10): 282- 290. doi: 10.6041/j.issn.1000-1298.2021.10.029
2	TONG J H, SHI H F, WU C Y, et al. Skewness correction and quality evaluation of plug seedling images based on Canny operator and Hough transform. Computers and Electronics in Agriculture, 2018, 155, 461- 472. doi: 10.1016/j.compag.2018.10.035
3	刘芳, 刘玉坤, 张白. 基于D-YOLOv3检测网络的温室叶菜幼苗图像检测. 江苏农业学报, 2021, 37(5): 1262- 1269. doi: 10.3969/j.issn.1000-4440.2021.05.022
	LIU F, LIU Y K, ZHANG B. Image detection of leafy vegetable seedlings in greenhouse based on D-YOLOv3 detection network. Jiangsu Journal of Agricultural Sciences, 2021, 37(5): 1262- 1269. doi: 10.3969/j.issn.1000-4440.2021.05.022
4	宋磊, 李嵘, 焦义涛, 等. 基于ResNeXt单目深度估计的幼苗植株高度测量方法. 农业工程学报, 2022, 38(3): 155- 163. URL
	SONG L, LI R, JIAO Y T, et al. Method for measuring seedling height based on ResNeXt monocular depth estimation. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(3): 155- 163. URL
5	邵明月, 张建华, 冯全, 等. 深度学习在植物叶部病害检测与识别的研究进展. 智慧农业(中英文), 2022, 4(1): 29- 46. URL
	SHAO M Y, ZHANG J H, FENG Q, et al. Research progress of deep learning in detection and recognition of plant leaf diseases. Smart Agriculture, 2022, 4(1): 29- 46. URL
6	SHARIF M, KHAN M A, IQBAL Z, et al. Detection and classification of citrus diseases in agriculture based on optimized weighted segmentation and feature selection. Computers and Electronics in Agriculture, 2018, 150, 220- 234. doi: 10.1016/j.compag.2018.04.023
7	张秀花, 静茂凯, 袁永伟, 等. 基于改进YOLOv3-Tiny的番茄苗分级检测. 农业工程学报, 2022, 38(1): 221- 229. URL
	ZHANG X H, JING M K, YUAN Y W, et al. Tomato seedling classification detection using improved YOLOv3-Tiny. Transactions of the Chinese Society of Agricultural Engineering, 2022, 38(1): 221- 229. URL
8	TONG J, YU J, WU C, et al. Health information acquisition and position calculation of plug seedling in greenhouse seedling bed. Computers and Electronics in Agriculture, 2021, 185, 106146. doi: 10.1016/j.compag.2021.106146
9	贺磊盈, 蔡丽苑, 武传宇. 基于机器视觉的幼苗自动嫁接参数提取. 农业工程学报, 2013, 29(24): 190- 195. doi: 10.3969/j.issn.1002-6819.2013.24.025
	HE L Y, CAI L Y, WU C Y. Vision-based parameters extraction of seedlings for grafting robot. Transactions of the Chinese Society of Agricultural Engineering, 2013, 29(24): 190- 195. doi: 10.3969/j.issn.1002-6819.2013.24.025
10	苗中华, 余孝有, 徐美红, 等. 基于图像处理多算法融合的杂草检测方法及试验. 智慧农业(中英文), 2020, 2(4): 103- 115. URL
	MIAO Z H, YU X Y, XU M H, et al. Automatic weed detection method based on fusion of multiple image processing algorithms. Smart Agriculture, 2020, 2(4): 103- 115. URL
11	ZHANG G, WEN Y S, TAN Y Z, et al. Identification of cabbage seedling defects in a fast automatic transplanter based on the maxIOU algorithm. Agronomy, 2020, 10(1): 65- 72. doi: 10.3390/agronomy10010065
12	王爱臣, 张敏, 刘青山, 等. 基于区域生长均值漂移聚类的苗期作物行提取方法. 农业工程学报, 2021, 37(19): 202- 210. doi: 10.11975/j.issn.1002-6819.2021.19.023
	WANG A C, ZHANG M, LIU Q S, et al. Seedling crop row extraction method based on regional growth and mean shift clustering. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37(19): 202- 210. doi: 10.11975/j.issn.1002-6819.2021.19.023
13	施玮囡. 基于深度学习的植物工厂幼苗表型信息获取及定植装备研究[D]. 杭州: 浙江大学, 2019.
	SHI W N. Research on phenotypic information acquisition and colonization equipment of plant factory seedlings based on deep learning[D]. Hangzhou: Zhejiang University, 2019. (in Chinese)
14	王璨, 武新慧, 张燕青, 等. 基于双注意力语义分割网络的田间苗期玉米识别与分割. 农业工程学报, 2021, 37(9): 211- 221. URL
	WANG C, WU X H, ZHANG Y Q, et al. Recognition and segmentation of maize seedlings in field based on dual attention semantic segmentation network. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37(9): 211- 221. URL
15	PATHAN M, PATEL N, YAGNIK H, et al. Artificial cognition for applications in smart agriculture: a comprehensive review. Artificial Intelligence in Agriculture, 2020, 4, 81- 95.
16	LI W, LIU C, CHEN M H, et al. A lightweight semantic segmentation model of Wucai seedlings based on attention mechanism. Photonics, 2022, 9(6): 393- 403.
17	GONG L, DU X F, ZHU K, et al. Pixel level segmentation of early-stage in-bag rice root for its architecture analysis. Computers and Electronics in Agriculture, 2021, 12(3): 186- 193.
18	娄路, 吕惠, 宋然. 基于多视角时间序列图像的植物叶片分割与特征提取. 农业机械学报, 2022, 53(1): 253- 260. URL
	LOU L, LÜ H, SONG R. Segmentation of plant leaves and features extraction based on muti-view and time-series image. Transactions of the Chinese Society for Agricultural Machinery, 2022, 53(1): 253- 260. URL
19	魏超宇, 韩文, 庞程, 等. 基于多尺度特征融合和密集连接网络的疏果期黄花梨植株图像分割. 江苏农业学报, 2021, 37(4): 990- 997. URL
	WEI C Y, HAN W, PANG C, et al. Image segmentation of Huanghua pear plants at fruit-thinning stage based on multi-scale feature fusion and dense connection network. Jiangsu Journal of Agricultural Sciences, 2021, 37(4): 990- 997. URL
20	庄前伟, 王志明, 吴龙贻, 等. 基于改进SOLOv2的穴盘幼苗图像分割方法. 南京农业大学学报, 2023, 46(1): 200- 209. URL
	ZHUANG Q W, WANG Z M, WU L Y, et al. Image segmentation method of plug seedlings based on improved SOLOv2. Journal of Nanjing Agricultural University, 2023, 46(1): 200- 209. URL
21	王祥, 舒军, 雷建军, 等. 基于改进YOLACT++的碧根果图像实例分割模型. 中南民族大学学报(自然科学版), 2022, 41(5): 613- 622. URL
	WANG X, SHU J, LEI J J, et al. Instance segmentation model of pecan image based on improved YOLACT++. Journal of South-Central University for Nationalities (Natural Science Edition), 2022, 41(5): 613- 622. URL
22	BOLYA D, ZHOU C, XIAO F Y, et al. YOLACT: real-time instance segmentation[C]//Proceedings of IEEE/CVF International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2020: 9156-9165.
23	QIAO S Y, CHEN L C, YUILLE A. DetectoRS: detecting objects with recursive feature pyramid and switchable atrous convolution[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2021: 10208-10219.
24	CHEN L C, ZHU Y K, PAPANDREOU G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation[C]//Proceedings of the 15th European Conference on Computer Vision. Berlin, Germany: Springer, 2018, 11211: 833-851.
25	XIE S N, GIRSHICK R, DOLLÁR P, et al. Aggregated residual transformations for deep neural networks[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2017: 5987-5995.
26	尚庆茂, 司军. 绿叶类蔬菜集约化育苗技术规程. 中国蔬菜, 2022,(9): 100- 105. URL
	SHANG Q M, SI J. Technical regulations for intensive seeding cultivation of green leafy vegetables. China Vegetables, 2022,(9): 100- 105. URL
27	HE K M, GKIOXARI G, DOLLÁR P, et al. Mask R-CNN[C]//Proceedings of IEEE International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2017: 2980-2988.
28	CHEN L C, PAPANDREOU G, KOKKINOS I, et al. DeepLab: semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 40(4): 834- 848.
29	TIAN Z, SHEN C H, CHEN H, et al. FCOS: fully convolutional one-stage object detection[C]//Proceedings of IEEE/CVF International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2020: 9626-9635.
30	WANG X, KONG T, SHEN C, et al. Solo: segmenting objects by locations[C]//Proceedings of European Conference on Computer Vision. Berlin, Germany: Springer, 2020: 649-665.
31	FANG Y X, YANG S S, WANG X G, et al. Instances as queries[C]//Proceedings of IEEE/CVF International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2022: 6890-6899.

[1]	马素刚, 陈期梅, 侯志强, 杨小宝, 张子贤. 基于密集连接与特征增强的语义分割算法[J]. 计算机工程, 2023, 49(3): 263-270.
[2]	窦允冲, 侯进, 曾雷鸣, 陈子锐. 基于反馈机制与空洞卷积的道路小目标检测网络[J]. 计算机工程, 2023, 49(1): 287-294.
[3]	徐鹏, 李勇, 金德鹏, 苏厉, 曾烈光. 改进的两阶段虚拟网映射算法[J]. 计算机工程, 2012, 38(5): 79-82.
[4]	金倩倩, 林丹. 求解UCARPP问题的变邻域搜索算法[J]. 计算机工程, 2012, 38(21): 290-292.
[5]	邹小林. 一种改进的最小最大割算法[J]. 计算机工程, 2012, 38(15): 215-217,221.
[6]	常志军, 杨鑫. 一种生物在体荧光成像的自适应分割算法[J]. 计算机工程, 2011, 37(4): 218-220.
[7]	眭博聪, 陈敏刚, 高岩, 马利庄. 基于三维体的多分辨率视频分割算法[J]. 计算机工程, 2011, 37(20): 282-284.
[8]	陈青青. 递归随机分割算法在考试系统中的应用[J]. 计算机工程, 2011, 37(14): 268-270.
[9]	黄歆;桑楠. 基于DOM树和递归X-Y分割算法的Zone树模型[J]. 计算机工程, 2009, 35(5): 53-55.
[10]	王小康;杨明. 具有容侵能力的多播密钥管理方案[J]. 计算机工程, 2008, 34(20): 190-192.

选择文件类型/文献管理软件名称

选择包含的内容

基于YOLACT-RFX模型的穴盘甘蓝苗株分割算法

Segmentation Algorithm of Plug Cabbage Seedlings Based on YOLACT-RFX Model

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 16

参考文献 31

相关文章 10

编辑推荐

Metrics

本文评价

模态框（Modal）标题

选择文件类型/文献管理软件名称

选择包含的内容

基于YOLACT-RFX模型的穴盘甘蓝苗株分割算法

Segmentation Algorithm of Plug Cabbage Seedlings Based on YOLACT-RFX Model

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 16

参考文献 31

相关文章 10

编辑推荐

Metrics

本文评价