AUTHOR=Li Rujia , He Yiting , Li Yadong , Qin Weibo , Abbas Arzlan , Ji Rongbiao , Li Shuang , Wu Yehui , Sun Xiaohai , Yang Jianping 

TITLE=Identification of cotton pest and disease based on CFNet- VoV-GCSP -LSKNet-YOLOv8s: a new era of precision agriculture

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 15 - 2024

YEAR=2024

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2024.1348402

DOI=10.3389/fpls.2024.1348402

ISSN=1664-462X

ABSTRACT=Addressing the challenges posed by traditional methods in detecting cotton leaf pests and diseases under natural conditions, this study proposes a cotton pest and disease identification method based on CFNet- VoV-GCSP -LSKNet- YOLOv8s. Built upon YOLOv8s, the model incorporates the CFNet module to replace all C2F modules in the backbone network, thereby enhancing the network's capability for multi-scale object feature fusion. Subsequently, the VoV-GCSP module replaces all C2F modules in the YOLOv8s head, maintaining model accuracy while reducing computational burden. In addition, the LSKNet attention mechanism is integrated into the small object layers of both the backbone network and the head to improve small object detection performance. Lastly, the model's convergence performance is enhanced by introducing the XIoU loss function. Experimental results demonstrate that the proposed method achieves a precision (P) of 89.9%, a recall rate (R) of 90.7%, and a mean average precision (mAP@0.5) of 93.7%, with a model memory footprint of 23.3MB. and the detection time is 8.01ms. Compared with YOLO v5s, YOLOX, YOLO v7, Faster R-CNN, YOLOv8n, YOLOv7-tiny, CenterNet, EfficientDet and YOLOv8s, the average accuracy is improved by 2.5%, 5%, 12.5%, 9%, 3.7%, 21.8% respectively, 4.9%, 12.1% and 1.2%. This indicates that the present method can effectively identify cotton pests and diseases in complex environment, providing a valuable technical resource for the identification and control of cotton pests and diseases.