Study on the Detection of Defoliation Effect of an Improved YOLOv5x Cotton

In order to study the detection effect of cotton boll opening after spraying defoliant, and solve problem low efficiency traditional manual methods for use this proposed a method improved YOLOv5x+ algorithm. Convolution Attention Module (CBAM) was embedded Conv enhance network’s feature extraction ability, suppress background information interference, enable network focus better on targets in process. At same time, depth separable convolution (DWConv) used replace ordinary (Conv) YOLOv5x model, reducing kernel parameters algorithm, amount calculation, improving speed Finally, layer added make algorithm have higher accuracy detecting small size cotton. The test results show that rate P (%), recall R mAP value (%) reach 90.95, 89.16, 78.47 respectively, which are 8.58, 8.84, 5.15 than convergence is faster, error smaller, resolution target improved, can meet defoliant.