A Pest Detection Algorithm Based on Improved YOLO

A pest detection algorithm based on improved YOLO is proposed in this paper to improve the recognition rate of pest detection. Based on the original YOLOv5 model, a new point distance loss function is proposed to reduce redundant computations and improve the model performance. Then an attention module is added to the model to reduce the complexity, keep the performance and improve the detection recognition rate. The experiments on the test set verified the effectiveness of the proposed model. The results show that the mean Average Precision of our proposed model is 95.9% and the mean detection time is 6.1 ms. The proposed model keeps the lightweight features of YOLOv5 with good prospects in smart agriculture applications.