王玲,张旗,冯天赐,王一博,李雨桐,陈度.基于YOLO v7-ST模型的小麦籽粒计数方法研究[J].农业机械学报,2023,54(10):188-197,204.
WANG Ling,ZHANG Qi,FENG Tianci,WANG Yibo,LI Yutong,CHEN Du.Wheat Grain Counting Method Based on YOLO v7-ST Model[J].Transactions of the Chinese Society for Agricultural Machinery,2023,54(10):188-197,204.
|
摘要点击次数: 349
全文下载次数: 198
|
| 基于YOLO v7-ST模型的小麦籽粒计数方法研究 [下载全文] |
| Wheat Grain Counting Method Based on YOLO v7-ST Model [Download Pdf][in English] |
| 投稿时间：2022-12-02 |
| DOI：10.6041/j.issn.1000-1298.2023.10.018 |
| 中文关键词: 小麦 籽粒计数 目标检测 离散度等级 电磁振动 YOLO v7-ST |
| 基金项目:中国农业大学横向课题项目（69193028） |
|
| 中文摘要:针对小麦考种过程中籽粒堆积、粘连和遮挡现象导致计数准确率低等问题，本文基于电磁振动原理设计了高通量小麦籽粒振动分离装置，通过分析受力探讨了籽粒离散分离程度的主要影响因素，并引入二阶离散系数建立了籽粒离散度等级评价方法。在此基础上，引入Swin Transformer模块构建YOLO v7-ST模型，对不同离散度等级下小麦籽粒进行计数性能测试。试验结果表明，YOLO v7-ST模型在3种离散度等级下平均计数准确率、F1值和平均计数时间的总平均值分别为99.16%、93%和1.19s，相较于YOLO v7、YOLO v5和Faster R-CNN模型，平均计数准确率分别提高1.03、2.34、15.44个百分点，模型综合评价指标F1值分别提高2、3、16个百分点，平均计数时间较YOLO v5和Faster R-CNN分别减少0.41s和0.36s，仅比YOLO v7模型增大0.09s。因此，YOLO v7-ST模型可实现多种离散度等级下不同程度籽粒遮挡和粘连问题的准确快速检测，大幅提高小麦考种效率。 |
| WANG Ling ZHANG Qi FENG Tianci WANG Yibo LI Yutong CHEN Du |
| China Agricultural University;Heilongjiang Agricultural Machinery Service Co., Ltd., Beidahuang Agricultural Service Group Co., Ltd. |
| Key Words:wheat grain count object detection dispersion level electromagnetic vibration YOLO v7-ST |
| Abstract:Aiming at the problems of low counting accuracy due to seed accumulation, sticking and shading phenomena in the wheat seed testing process, a high-throughput wheat seed vibration separation device was designed based on the principle of electromagnetic vibration. The main influencing factors of the degree of seed dispersion and separation were discussed by analyzing the forces, and the second order dispersion coefficient was introduced to establish the seed dispersion grade evaluation method. On this basis, the YOLO v7-ST model was then built by using the Swin Transformer module and was tested for counting performance under different discrete degree levels. The experimental results showed that the mean counting accuracy, F1 value and mean counting time of the YOLO v7-ST model were 99.16%, 93% and 1.19s under the three dispersion levels, respectively. Compared with that of the YOLO v7, YOLO v5 and Faster R-CNN models, the mean counting accuracy was improved by 1.03 percentage points, 2.34 percentage points and 15.44 percentage points, respectively, and the F1 values of the comprehensive evaluation index of the model was increased by 2 percentage points, 3 percentage points and 16 percentage points, respectively. The mean counting time was decreased by 0.41s and 0.36s compared with that of YOLO v5 and Faster R-CNN, respectively, and it was only 0.09s slower than that of the YOLO v7 model. Overall, the YOLO v7-ST model provided accurate and efficient detection of grains under various discrete degree levels, significantly improved the efficiency of wheat breeding. |
Transactions of the Chinese Society for Agriculture Machinery (CSAM), in charged of China Association for Science and Technology (CAST), sponsored by CSAM and Chinese Academy of Agricultural Mechanization Science(CAAMS), started publication in 1957. It is the earliest interdisciplinary journal in Chinese which combines agricultural and engineering. It always closely grasps the development direction of agriculture engineering disciplines and the published papers represent the highest academic level of agriculture engineering in China. Currently, nearly 8,000 papers have been already published. There are around 3,000 papers contributed to the journal each year, but only around 600 of them will be accepted. Transactions of CSAM focuses on a wide range of agricultural machinery, irrigation, electronics, robotics, agro-products engineering, biological energy, agricultural structures and environment and more. Subjects in Transactions of the CSAM have been embodied by many internationally well-known index systems, such as: EI Compendex, CA, CSA, etc.
|
| 下载PDF阅读器 |
|
|
|
