An Improved Lightweight Common Object Detection Algorithm based on YOLOv11s

To address the need for integration capability of mobile robots, a research is proposed on the improved YOLO 11s object recognition ranging method. The YOLO 11s model is first combined with the C3k2 module by introducing the MSCV module to extract the features of the input dimensions at multiple scales, thus enhancing the target detection capability of the model; in the lightweighting of the model by introducing the LDCM module, which makes the model achieve lightweighting with minimal loss of accuracy; and finally, the localization capability of the model for the target is enhanced by introducing the PIoU v2 loss function. In comparison with the benchmark model, the accuracy P, mAP@50, and FPS of the improved YOLO 11s model increases by 2.9%, 1.81%, and 19.88%, respectively, and the overall parameters and complexity of the model are reduced by 22.75% and 7.98%, respectively, which not only realizes the lightweighting of the model, but also effectively enhances the model's detection capability.