Blueberry flower detection algorithm based on improved YOLOv8.

In order to realize the accurate and rapid identification of blueberry flowers in the natural environment, this study improved the structure of the YOLOv8 network and proposed an improved SPD-YOLOv8 blueberry flower detection algorithm based on YOLOv8. First, SPD-Conv is used instead of Conv in the Backbone section to improve the detection performance of the model; second, to improve the detection effect of the model on dense objects, replace the NMS with the Soft-NMS; then, for the loss of semantic information, the addition of small target detection layer and finally, replace the original loss function to improve the accuracy of target detection through a dynamic focusing mechanism to improve the performance of the model. The experimental results show that the accuracy (Precision) and mean accuracy (mAP) of the improved model in the blueberry flower data set are $9.5 \%$ and $4.8 \%$ higher respectively compared with the original model, and the model size is reduced by 0.2MB. Compared with the original YOLOv8 model, the model has better performance in detection.