An Improved Lightweight YOLOv5 Algorithm for Detecting Strawberry Diseases

This paper proposes an improved lightweight YOLOv5 model for the real-time detection of strawberry diseases. The ghost convolution (GhostConv) module is incorporated into network, reducing parameter numbers and floating-point operations (FLOPs) extracting feature information using backbone network. An involution operator utilized in network to expand receptive field, enhance spatial on disease characteristics, reduce number FLOPs model. A convolutional block attention (CBAM) network’s ability extract features suppress non-critical information. upsampling replaced by a called Content-Aware ReAssembly Features (CARAFE), which extracts map enhances focus features. experimental results open-source dataset show that achieves mean average precision (mAP)@0.5 94.7% with 3.9 M parameters 3.6 G FLOPs. has higher than original one lower hardware requirements, providing new strategy identification control.