Rice Disease Detection Method Based on Improved YOLOv8
Aiming at the problems of insufficient detection accuracy and high complexity of existing rice disease detection methods,an improved YOLOv8 method for rice disease detection is proposed.Firstly,the lightweight network GhostNet is introduced into the backbone network,and the C2fGhost module is constructed to replace the original C2f module,which reduces the number of parameters,floating point calculations and model size of the al-gorithm,and reduces the complexity of the algorithm.Secondly,the EMA attention mechanism is added to the neck network to enhance the ability of key information extraction.Finally,the NWD loss function is introduced to combine with the CIoU loss function to improve the detection accuracy of the algorithm.The experimental results show that compared with YOLOv8n,the overall average accuracy(mAP@0.5)of the improved algorithm GEN-YOLO is increased by 1.4 percentage points,the number of parameters is reduced by 0.452M,the amount of float-ing point calculation is reduced by 1.2G,and the model size is reduced by 0.826MB.The improved algorithm ef-fectively improves the detection accuracy while ensuring lightweight.Moreover,the proposed method is superior to other mainstream object detection methods in terms of detection accuracy and algorithm complexity,which indi-cates that the proposed method is advanced.
rice disease detectionYOLOv8GhostNetEMA attention moduleNWD loss function
NETL
NSTL
万方数据
