[1]吴康磊,金秀,饶元,等.基于虚拟数据和旋转目标检测分析的大豆豆荚表型参数测量方法[J].江苏农业学报,2024,(07):1245-1259.[doi:doi:10.3969/j.issn.1000-4440.2024.07.011]
　WU Kanglei,JIN Xiu,RAO Yuan,et al.Measurement method for soybean pod phenotypic parameters based on virtual data and rotated object detection analysis[J].,2024,(07):1245-1259.[doi:doi:10.3969/j.issn.1000-4440.2024.07.011]
点击复制

基于虚拟数据和旋转目标检测分析的大豆豆荚表型参数测量方法()
分享到：

江苏农业学报[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2024年07期
页码:
1245-1259
栏目:
农业信息工程
出版日期:
2024-07-30

文章信息/Info

Title:
Measurement method for soybean pod phenotypic parameters based on virtual data and rotated object detection analysis
作者:
吴康磊1; 2; 金秀1; 2; 饶元1; 2; 李佳佳3;  王晓波3; 王坦1; 2; 江朝晖1; 2
(1.安徽农业大学信息与人工智能学院,安徽合肥230036；2.农业农村部农业传感器重点实验室,安徽合肥230036;3.安徽农业大学农学院,安徽合肥230036)
Author(s):
WU Kanglei1; 2; JIN Xiu1; 2; RAO Yuan1; 2; LI Jiajia3; WANG Xiaobo3; WANG Tan1; 2; JIANG Zhaohui1; 2
(1.College of Information and Artificial Intelligence, Anhui Agricultural University, Hefei 230036, China;2.Key Laboratory of Agricultural Sensors, Ministry of Agriculture and Rural Affairs, Hefei 230036, China;3.College of Agronomy, Anhui Agricultural University, Hefei 230036, China)
关键词:
大豆考种; 豆荚表型; 虚拟数据; 旋转目标检测; YOLOv7-tiny
Keywords:
soybean seed evaluation; soybean pod phenotype; virtual data; rotated object detection; YOLOv7-tiny
分类号:
TP391.4
DOI:
doi:10.3969/j.issn.1000-4440.2024.07.011
文献标志码:
A
摘要:
为解决传统大豆考种过程中人工测量大豆豆荚表型参数耗时费力的问题以及现有的自动化测量方式存在的人工数据标注需求量大、环境适应能力弱、计算代价高等问题,本研究提出一种基于虚拟数据集生成和旋转目标检测分析的豆荚关键表型参数自动化测量方法,重点关注荚长和荚宽的测量。该方法基于YOLOv7-tiny提出一种改进的豆荚检测模型(CSL-YOLOv7-tiny),通过引入环形平滑标签使模型获得对旋转目标的检测能力,提升对无序摆放的狭长豆荚目标检测的质量。为避免人工标注训练数据,采用虚拟图像生成方法得到含标注信息的虚拟豆荚数据集和虚拟硬币与豆荚混合数据集。利用迁移学习策略,将模型从虚拟豆荚数据集迁移至虚拟硬币与豆荚混合数据集,积累模型对豆荚特征的提取能力。设计一种基于K-均值聚类的后处理方法,对检测到的旋转边界框进行分析，得到荚长和荚宽,以减少拍摄环境差异带来的测量误差。试验结果表明,在无任何训练数据标注的条件下,使用虚拟图像训练的CSL-YOLOv7-tiny对硬币和豆荚目标检测的最优mAP0.50和mAP0.50∶0.95分别达到了99.3%和78.0%,其模型大小和推理时间分别仅为12.92 MB和12.5 ms,荚长和荚宽测量的决定系数（R2）分别达到了0.94和0.86,与实际测量均值分别仅相差0.42 mm和0.02 mm。此外,通过对本研究提出的方法进行对比分析,验证了其在模型训练、轻量化部署以及不同考种环境适应能力上的优势。研究结果可为大豆豆荚表型参数的自动化、智能化测量系统的研发提供参考,为加速优质高产大豆的选育进程提供支撑。
Abstract:
To solve the problems such as time-consuming and labor-intensive of manual measurement for soybean pod phenotypic parameters in traditional soybean seed evaluation processes, as well as the large quantity demand for manual data annotation, weak environmental adaptation and high computational costs in existing automated measurement methods, an automated measurement method for pod key phenotypic parameters which was mainly focused on pod length and width measuring was proposed in this study, based on virtual dataset generation and rotated object detection analysis. An improved pod detection model (CSL-YOLOv7-tiny) was proposed by the method based on YOLOv7-tiny. The Circular Smooth Label was introduced to enable the model to obtain the capability for rotated object detection, and to improve the quality of detecting elongated pod targets in a disorganized arrangement. To avoid manual annotation of training data, virtual image generation method was used to get virtual pod dataset as well as virtual coin and pod mixture dataset containing annotation information. Transfer learning strategy was employed to transfer the model from the virtual pod dataset to the virtual coin and pod mixture dataset, which accumulated the model’s ability in pod features extracting. A post-processing method based on K-means clustering was designed to analyze the detected rotated bounding boxes, and obtained pod length and width, which reduced measurement errors caused by shooting environmental differences. Experimental results showed that under the condition of no training data annotation, CSL-YOLOv7-tiny trained by virtual images obtained the optimal mAP0.50 and mAP0.50∶0.95 for coin and pod targets detection, which were 99.3% and 78.0%, respectively. The model size and inference time were only 12.92 MB and 12.5 ms respectively, and the determination coefficients (R2) for pod length and width measurement reached 0.94 and 0.86 respectively, with only 0.42 mm and 0.02 mm differences compared with actual measurements. Furthermore, by comparative analysis of the proposed method, the advantages in model training, lightweight deployment and adaptation to different breeding environments were validated. The research results can provide reference for development of automated and intelligent measurement system of soybean pod phenotypic parameter and can support the acceleration of high-quality and high-yield soybean breeding.

参考文献/References:

[1]PADALKAR G, MANDLIK R, SUDHAKARAN S, et al. Necessity and challenges for exploration of nutritional potential of staple-food grade soybean[J]. Journal of Food Composition and Analysis,2023,117:105093.
[2]陈雨生,江一帆,张瑛. 中国大豆生产格局变化及其影响因素[J]. 经济地理,2022,42(3):87-94.
[3]宋晨旭,于翀宇,邢永超,等. 基于OpenCV的大豆籽粒多表型参数获取算法[J]. 农业工程学报,2022,38(20):156-163.
[4]XIANG S, WANG S Y, XU M, et al. YOLO POD:a fast and accurate multi-task model for dense Soybean Pod counting[J]. Plant Methods,2023,19(1):8.
[5]ZHOU W, CHEN Y J, LI W H, et al. SPP-extractor:automatic phenotype extraction for densely grown soybean plants[J]. The Crop Journal,2023,11(5):1569-1578.
[6]周华茂,王婧,殷华,等. 基于改进Mask R-CNN模型的秀珍菇表型参数自动测量方法[J]. 智慧农业,2023,5(4):117-126.
[7]CHEN S, ZOU X J, ZHOU X Z, et al. Study on fusion clustering and improved yolov5 algorithm based on multiple occlusion of Camellia oleifera fruit[J]. Computers and Electronics in Agriculture,2023,206:107706.
[8]RONG J C, ZHOU H, ZHANG F, et al. Tomato cluster detection and counting using improved YOLOv5 based on RGB-D fusion[J]. Computers and Electronics in Agriculture,2023,207:107741.
[9]UZAL L C, GRINBLAT G L, NAMíAS R, et al. Seed-per-pod estimation for plant breeding using deep learning[J]. Computers and Electronics in Agriculture,2018,150:196-204.
[10]闫壮壮, 闫学慧,石嘉, 等. 基于深度学习的大豆豆荚类别识别研究[J]. 作物学报,2020,46(11):1771-1779.
[11]郭瑞,于翀宇,贺红,等. 采用改进YOLOv4算法的大豆单株豆荚数检测方法[J]. 农业工程学报,2021,37(18):179-187.
[12]宁姗,陈海涛,赵秋多,等. 基于IM-SSD+ACO算法的整株大豆表型信息提取[J]. 农业机械学报,2021,52(12):182-190.
[13]王跃亭,王敏娟,孙石,等. 基于图像处理和聚类算法的待考种大豆主茎节数统计[J]. 农业机械学报,2020,51(12):229-237.
[14]张小斌,谢宝良,朱怡航,等. 基于图像处理技术的菜用大豆豆荚高通量表型采集与分析[J]. 核农学报,2022,36(3):602-612.
[15]赵岩,张人天,董春旺,等. 采用改进Unet网络的茶园导航路径识别方法[J]. 农业工程学报,2022,38(19):162-171.
[16]杨蜀秦,王帅,王鹏飞,等. 改进YOLOX检测单位面积麦穗[J]. 农业工程学报,2022,38(15):143-149.
[17]翔云,陈其军,宋栩杰,等. 基于深度学习的菜用大豆荚型表型识别方法[J]. 核农学报,2022,36(12):2391-2399.
[18]JIANG P Y, ERGU D, LIU F Y, et al. A review of Yolo algorithm developments[J]. Procedia Computer Science,2022,199:1066-1073.
[19]DIWAN T, ANIRUDH G, TEMBHURNE J V. Object detection using YOLO:challenges, architectural successors, datasets and applications[J]. Multimedia Tools and Applications,2023,82(6):9243-9275.
[20]LI S, YAN Z Z, GUO Y X, et al. SPM-IS:an auto-algorithm to acquire a mature soybean phenotype based on instance segmentation[J]. The Crop Journal,2022,10(5):1412-1423.
[21]JUNG Y, BYUN S, KIM B, et al. Harnessing synthetic data for enhanced detection of Pine Wilt Disease:an image classification approach[J]. Computers and Electronics in Agriculture,2024,218:108690.
[22]BARRIENTOS-ESPILLCO F, GASC E, LPEZ-GONZLEZ C I, et al. Semantic segmentation based on deep learning for the detection of Cyanobacterial harmful algal blooms (CyanoHABs) using synthetic images[J]. Applied Soft Computing,2023,141:110315.
[23]ABBAS A, JAIN S, GOUR M, et al. Tomato plant disease detection using transfer learning with C-GAN synthetic images[J]. Computers and Electronics in Agriculture,2021,187:106279.
[24]YANG S, ZHENG L H, YANG H J, et al. A synthetic datasets based instance segmentation network for high-throughput soybean pods phenotype investigation[J]. Expert Systems with Applications,2022,192:116403.
[25]WANG C Y, BOCHKOVSKIY A, LIAO H Y M. YOLOv7:trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]. Vancouver:IEEE,2023:7464-7475.
[26]YANG X, YAN J C. Arbitrary-oriented object detection with circular smooth label[C]. Glasgow:Springer,2020:677-694.
[27]GIRSHICK R. Fast r-cnn[C]. Santiago:IEEE,2015:1440-1448.

备注/Memo

备注/Memo:
收稿日期：2024-04-03基金项目：国家自然科学基金项目(32371993)；安徽省重点研究与开发计划项目(202204c06020026、2023n06020057)；安徽省高校自然科学研究重大项目(2022AH040125、 2023AH040135)作者简介：吴康磊(2003-),男,安徽蚌埠人,本科,研究方向为机器视觉和深度学习。(E-mail)wukanglei272@gmail.com通讯作者：饶元,(E-mail)raoyuan@ahau.edu.cn
更新日期/Last Update: 2024-09-14