Unmanned aerial vehicle detection method based on YOLO network lightweight deployment

The invention relates to an unmanned aerial vehicle detection method based on YOLO network lightweight deployment, belongs to the technical field of unmanned aerial vehicle detection, and is suitable for a YOLO model to carry out lightweight deployment acceleration on edge computing platforms such as Rayleigh RK3588 and the like so as to realize efficient target detection. The convolution and attention mechanism is improved by using the Ghost convolution and the decoupling full-connection attention mechanism, and the operation speed of the model is greatly improved while the model precision is kept. Through the improvement, the target detection task can be completed more quickly on the edge computing platform, and the application scene of unmanned aerial vehicle detection with the high real-time requirement is met. 本发明涉及一种基于YOLO网络轻量化部署的无人机检测方法，属于无人机检测技术领域，适用于YOLO模型在瑞芯微RK3588等边缘计算平台进行轻量化部署加速，实现高效的目标检测。通过使用Ghost卷积和解耦全连接注意力机制改进卷积和注意力机制，本发明在保持模型精度的同时，大幅提升了模型的运行速度。这种改进使得目标检测任务在边缘计算平台上能够更快速地完成，满足了无人机检测这种实时性要求较高的应用场