摘要
针对采摘机器人在实际应用中所需要的视觉检测能力,提出一种基于深度学习的移动端柑橘检测算法,实现对柑橘类水果的快速实时检测。通过PyTorch框架将YOLOv5s、YOLOv5s6、YOLOv5m等模型在自制的柑橘数据集下进行训练,将训练好的模型转化为ONNX模型,再将ONNX模型转化为NCNN模型,并进行Android移动端部署。试验结果表明:NCNN深度学习框架下的YOLOv5s6识别率达到92.9%,召回率达到75.7%,能较好地满足采摘机器人实际应用需求。
In view of the visual detection ability required by the picking robot in practical application,a mobile citrus detection algorithm based on deep learning was proposed to realize the rapid real-time detection of citrus fruits.Models such as YOLOv5s,YOLOv5s6,and YOLOv5m were trained on self-made citrus datasets through the PyTorch framework,and the trained models were transformed into ONNX models,and the ONNX models were converted into NCNN models,and deployed on Android mobile terminals.The experimental results show that the recognition rate of YOLOv5s6 under the NCNN deep learning framework reaches 92.9%and the recall rate reaches 75.7%,which can better meet the practical application needs of picking robots.
作者
刘学
郑哲文
朱洪前
LIU Xue;ZHENG Zhe-wen;ZHU Hong-qian(School of Materials Science and Engineering,Central South University of Forestry and Technology,Changsha Hunan 410004)
出处
《长沙航空职业技术学院学报》
2024年第3期27-31,共5页
Journal of Changsha Aeronautical Vocational and Technical College
