A Lightweight Detection Algorithm YOLOv8-L Based on YOLOv8n

With the rise of intelligent detection technology, object detection has become the key to real-time monitoring. However, due to the large model size and high computational requirements, traditional algorithms are often difficult to deploy on resource constrained devices. To address the issue of lightweighting, this article introduces the lightweight model YOLOv8-L based on YOLOv8n. YOLOv8-L replaced the Conv module with an improved GhostShuffie module and partially replaced C2f with the C3Ghost module, significantly reducing parameters and computational complexity. In addition, the proposed CgfmEma module ensures strong detection accuracy through multi-scale feature guidance.Extensive experiments on different datasets have confirmed the generalization ability of YOLOv8-L. Compared with YOLOv8n, YOLOv8-L reduces parameters by about 30% and computational complexity by about 14.6%, while maintaining comparable mAP50 scores. These results highlight that YOLOv8-L is an efficient and deployable solution for real-time object detection in various scenarios.