SCM-YOLO: a lightweight object detection network for real-time UAVs deployment

At display, unmanned airborne vehicles (UAVs) play an increasingly important role in different areas of human life, but a considerable number of tasks put forward higher and higher requirements for real-time and accuracy of object detection in the scenario of UAVs. To address the trade-off between detection accuracy and processing speed faced by unmanned aerial vehicle real-time object detection due to limited computing resources. This paper proposes a detection framework based on YOLOv11: SCM-YOLO. This framework replaces the original backbone network with ShuffleNetV2 and introduces two key innovations: Cross-Scale Feature Fusion Module (CCFM) and Mixed Local Channel Attention (MLCA). This improvement resulted in a 63% reduction in parameter count for SCM-YOLO compared to the original YOLOv11n model, only 0.9M, while mAP (%) increased by 1.18% Thus, the real-time object detection of UAVS equipment can be realized by SCM-YOLO.