Light Weight Apple Leaf Disease Detection Method Based on Improved YOLOv8

In order to improve the accuracy of apple leaf disease detection, reduce the number of model parameters, realize lightweight module, and make its deployment more convenient. this study, an improved YOLOv8n method for the detection of light weight apple leaf disease was proposed. Firstly, based on Partial Convolution (PConv), a new CSPPC is proposed to replace C2f in the network, reducing the number of model parameters and achieving lightweight modules. Secondly, CBAM (Convolutional Block Attention Module) is added to the convolutional block attention Module to strengthen the semantic and positional information in the features, improve the feature fusion ability of the model, and improve the detection accuracy of apple leaf diseases. Finally, Wise-IoU (weighted interpolation of sequential evidence for intersection over union) boundary loss function is used to replace the original loss function. Improve the network boundary box regression performance and small target disease detection effect. The test results showed that for the apple leaf disease data set collected under natural environment, the parameter number and model size of the improved model were reduced by 26.1 and 25.6 percentage points, respectively, compared with the original YOLOv8n baseline model. At the same time, the precision and mean average precision (mAP) are increased by 1.0 and 0.9 percentage points respectively compared with the original model, which is better than other comparison target detection algorithms. It can provide reference for the deployment.