[Objectives]The paper was to explore a faster and more accurate detection method for citrus psyllid to prevent and control yellow-shoot disease and inhibit its transmission.[Methods]We used an improved YOLOX based edg...[Objectives]The paper was to explore a faster and more accurate detection method for citrus psyllid to prevent and control yellow-shoot disease and inhibit its transmission.[Methods]We used an improved YOLOX based edge detection method for psyllid,added Convolutional Block Attention Module(CBAM)to the backbone network,and further extracted important features in the channel and space dimensions.The Cross Entropy Loss in the object loss was changed to Focal Loss to further reduce the missed detection rate.[Results]The algorithm described in the study fitted in with the detection platform of psyllid.The data set of psyllid was taken in Lianjiang Orange Garden,Zhanjiang City,Guangdong Province,deeply adapted to the actual needs of agricultural and rural development.Based on YOLOX model,the backbone network and loss function were improved to achieve a more excellent detection method of citrus psyllid.The AP value of 85.66%was obtained on the data set of citrus psyllid,which was 2.70%higher than that of the original model,and the detection accuracies were 8.61%,4.32%and 3.62%higher than that of YOLOv3,YOLOv4-Tiny and YOLOv5-s,respectively,which had been greatly improved.[Conclusions]The improved YOLOX model can better identify citrus psyllid,and the accuracy rate has been improved,laying a foundation for the subsequent real-time detection platform.展开更多
基金Supported by Research and Development Program in Key Areas of Guangdong Province(2020B0202090005)Lianjiang Think Tank Enterprise Project"Demonstration of Intelligent Monitoring and Ecological Prevention and Control Technology of Red Orange Yellow-shoot Disease and Psyllid in Lianjiang"。
文摘[Objectives]The paper was to explore a faster and more accurate detection method for citrus psyllid to prevent and control yellow-shoot disease and inhibit its transmission.[Methods]We used an improved YOLOX based edge detection method for psyllid,added Convolutional Block Attention Module(CBAM)to the backbone network,and further extracted important features in the channel and space dimensions.The Cross Entropy Loss in the object loss was changed to Focal Loss to further reduce the missed detection rate.[Results]The algorithm described in the study fitted in with the detection platform of psyllid.The data set of psyllid was taken in Lianjiang Orange Garden,Zhanjiang City,Guangdong Province,deeply adapted to the actual needs of agricultural and rural development.Based on YOLOX model,the backbone network and loss function were improved to achieve a more excellent detection method of citrus psyllid.The AP value of 85.66%was obtained on the data set of citrus psyllid,which was 2.70%higher than that of the original model,and the detection accuracies were 8.61%,4.32%and 3.62%higher than that of YOLOv3,YOLOv4-Tiny and YOLOv5-s,respectively,which had been greatly improved.[Conclusions]The improved YOLOX model can better identify citrus psyllid,and the accuracy rate has been improved,laying a foundation for the subsequent real-time detection platform.
文摘为提高图像异型波形提取效率及准确性,提出基于卷积神经网络(Convolutional Neural Network,CNN)模型的图像快速定位处理方法,通过优化YOLOX模型结构对新建模型进行训练、测试及验证。结果表明:CP-YOLOX-48和CP-YOLOX-64两个模型验证集损失值较低且差异性较小,然而CP-YOLOX-64的浮点运算数比CP-YOLOX-48的运算值高了近1倍,故确定CP-YOLOX-48为最佳模型。与两种原始YOLOX模型相比,CP-YOLOX模型的图像处理效率、精准率(Accuracy)、召回率(Recall)及mAP(mean Average Precision)值均略高,且精准率大于90%,证实了CP-YOLOX-48模型具有较高的预测精度及提取效率。
