首页|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

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

扫码查看
原文链接
  • 万方数据
  • 维普
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.
Measurement method for soybean pod phenotypic parameters based on virtual data and rotated object detection analysis
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 de-mand for manual data annotation,weak environmental ad-aptation and high computational costs in existing automated measurement methods,an automated measurement method for pod key phenotypic parameters which was mainly fo-cused 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 en-able the model to obtain the capability for rotated object detection,and to improve the quality of detecting elongated pod tar-gets 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 de-signed to analyze the detected rotated bounding boxes,and obtained pod length and width,which reduced measurement er-rors 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 tar-gets 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 compara-tive 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 intelli-gent measurement system of soybean pod phenotypic parameter and can support the acceleration of high-quality and high-yield soybean breeding.

soybean seed evaluationsoybean pod phenotypevirtual datarotated object detectionYOLOv7-tiny

吴康磊、金秀、饶元、李佳佳、王晓波、王坦、江朝晖

展开 >

安徽农业大学信息与人工智能学院,安徽 合肥 230036

农业农村部农业传感器重点实验室,安徽 合肥 230036

安徽农业大学农学院,安徽 合肥 230036

大豆考种 豆荚表型 虚拟数据 旋转目标检测 YOLOv7-tiny

国家自然科学基金项目安徽省重点研究与开发计划项目安徽省重点研究与开发计划项目安徽省高校自然科学研究重大项目安徽省高校自然科学研究重大项目

32371993202204c060200262023n060200572022AH0401252023AH040135

2024

10.3969/j.issn.1000-4440.2024.07.011

江苏农业学报
江苏省农业科学院

江苏农业学报

CSTPCD北大核心
影响因子：1.093
ISSN：1000-4440
年,卷(期)：2024.40(7)