中国农机化学报2026，Vol.47Issue(5)：69-74,6.DOI:10.13733/j.jcam.issn.2095-5553.2026.05.009
基于深度学习的全自动穴盘苗分级系统设计
Design of fully automatic seedling classification system based on deep learning
摘要
Abstract
Considering the low efficiency of current tray seedling grading operations and the low accuracy of image classification,this paper proposes a fully automatic tray seedling grading system based on deep learning algorithms.It uses Siemens PLC as the controller and a three-axis robotic arm as the sorting executive mechanism.The system employs data augmentation methods to expand the collected tray seedling detection dataset,increasing the amount of training samples.It introduces the deep learning ResNet50 algorithm model,trains the model to obtain the optimal detection parameters,and performs virtual simulation and prototype verification on the proposed fully automatic tray seedling grading system.The test results show that the tray seedling grading system can effectively complete the transportation,recognition,and sorting of seedlings.The model's detection precision is 98.72%,the recall rate is 96.36%,the inference time is within 35 ms,and the intelligent sorting success rate is above 95%,verifying the effectiveness and feasibility of the fully automatic tray seedling sorting system,with stronger generalization and robustness.关键词
深度学习/穴盘苗/机器视觉/分级系统Key words
deep learning/seedling trays/machine vision/grading system分类
农业科技引用本文复制引用
孙璧文,高菊玲,胡程磊,王宜雷..基于深度学习的全自动穴盘苗分级系统设计[J].中国农机化学报,2026,47(5):69-74,6.基金项目
江苏农林职业技术学院科技项目(2023kj13) （2023kj13）
江苏农林职业技术学院大学生创新创业训练计划项目(202013103074Y) （202013103074Y）
江苏省第六期"333 高层次人才培养工程"项目((2022)3-23-070) （(2022）
