Ground Target Recognition Technology of UAV Based on YOLOv10

Drone ground target recognition technology is crucial in both military and civilian fields, yet it faces complex environmental challenges. This paper focuses on drone ground target recognition technology based on YOLOv10, detailing its current development status and comparing traditional methods with deep learning approaches. By standardizing data collection and using GANs to expand the dataset, this paper thoroughly analyzes the optimization design of each part of the YOLOv10 network structure. Experimental results show that metrics such as accuracy, recall rate, and mean average precision have improved during model training, with stable performance after stabilization. Further training using StyleGAN-generated samples can enhance model performance, leading to significant improvements in accuracy, recall rate, and mean average precision, providing an effective solution for the development of drone ground target recognition technology.