Accurate Detection Algorithm of Diaphorina Citri Based on YOLOv10s-C2M

Huanglong disease is a devastating disease that seriously threatens the safety of the citrus industry, and the accurate detection of Diaphorina Citri, as its main vector, is crucial for prevention and control. Aiming at the challenges of Diaphorina Citri, such as small size, easy background interference, complex orchard environment, and limited resources for deploying equipment, this paper proposes a novel lightweight composite attention module, MBCAM, and innovatively integrates it with the SE attention mechanism and the adaptive downsampling module (ADown), and introduces it into the YOLOv10s architecture, to construct a set of lightweight and highly efficient Diaphorina Citri detection model, YOLOv10s-C2M. Compared with the benchmark model, YOLOv10s-C2M has improved inference speed by 25.50%, reaching 3.8ms, while achieving a significant breakthrough in accuracy, mAP@0.5 improved by 4.53%, reaching 89.77%.This method balances detection accuracy and deployment efficiency, demonstrating good adaptability and practicality in actual citrus orchard scenarios. This study not only provides technical support for the intelligent monitoring of Diaphorina Citri, but also provides new ideas for the design of lightweight pest detection models, which has important agricultural intelligent application value and promotion prospects.