Detection of Litchi Leaf Diseases and Insect Pests Based on Improved FCOS

Litchi leaf diseases and pests can lead to issues such as a decreased yield, reduced fruit quality, farmer income. In this study, we aimed explore real-time accurate method for identifying pests. We selected three different orchards field investigation identified five common (Litchi mite, sooty mold, anthracnose, Mayetiola sp., algal spot) our research objects. Finally, proposed an improved fully convolutional one-stage object detection (FCOS) network disease pest detection, called FCOS Litch (FCOS-FL). The employs G-GhostNet-3.2 the backbone achieve model that is lightweight. central moment pooling attention (CMPA) mechanism introduced enhance features of addition, center sampling loss are by utilizing width height information real target, which effectively improves model’s generalization performance. propose localization function accuracy in detection. According characteristics small target pests, structure was redesigned improve effect targets. FCOS-FL has 91.3% (intersection over union (IoU) = 0.5) images types rate 62.0/ms, parameter size 17.65 M. Among them, sp. spot, difficult detect, reached 93.2% 92%, respectively. rapidly accurately detect leaf. outcome suitable deployment on embedded devices with limited resources mobile terminals, contribute achieving precise identification providing technical support diseases’ pests’ prevention control.