Accurate Detection and Precision Spraying of Corn and Weeds Using the Improved YOLOv5 Model

The precise identification of corn and weeds plays an important role in spraying. This paper proposed a lightweight model based on the improved yolov5s built precision spraying robot. Firstly, we used data augmentation method category balance agronomic characteristics to solve imbalance problem. Then, compared with yolov5s, yolov5l, yolov5m, yolov5x, found that has both real-time accuracy is easier deploy edge devices. Through feature map visualization experiment, extraction network can’t pay close attention target suppress noise. Therefore, added mechanism. In order improve performance model, designed C3-Ghost-bottleneck module. Finally, Compared original value [email protected] increased by 3.2%, file reduced 3.6 MB, AP for from 93.2% 96.3%, 85.6% 88.9%. experiment was carried out. recognition 83%, probability robot correctly identifying accurately 81%, detection speed 30ms/f. experimental results verify feasibility weeding effectiveness which can provide reference engineering application weeding.