XRXuefeng Rao

Xuefeng Rao

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Xuefeng Rao has a total of 10 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    A Lightweight UAV Small-Object Detection Algorithm Combining Background Suppression and Multiscale Feature Enhancement

    by Xiangmeng Pan, Xuefeng Rao, Biao Li, Juezhe Wang, Xiangnai Pan, Yu Wang, Chunmei Li, Yujie Zhao, Xi Huang

    This paper introduces the ABMS-YOLO algorithm to tackle challenges in small-object detection within UAV aerial photography, such as significant background interference, diverse object sizes, and limited edge computing capacity. Using YOLO11n as a foundation, the algorithm integrates background suppression, dynamic multi-scale feature enhancement, and lightweight end-to-end collaboration. The ABMS-YOLO algorithm consists of three main modules: the Background Suppression Module (BSM), which reduces background noise while maintaining small-object details through channel reorganization and spatial pyramid pooling; the Multi-Scale Feature Module (MSFM), which employs scale-aware dynamic dilated convolutions and adaptive receptive field adjustments to enhance multi-scale feature representation; and the Streamlined Lightweight Collaboration Module (SLCM), which achieves efficient processing via parameter-free spatial rearrangement and parameter-based channel filtering. Experimental results on datasets such as VisDrone2019, TinyPerson, and DOTAv1.0 reveal that ABMS-YOLO, with only 3.04 million parameters—a 17.8% increase over the baseline—improves mAP50 by 12.9%, 6.6%, and 5.8% to 45.1%, 19%, and 44.3%, respectively, while maintaining an FPS of 231.38. The algorithm’s overall detection performance surpasses both contemporary mainstream and representative algorithms, fulfilling the demands for real-time edge detection.

  • Open Access

    Dual-Scale Parallel Training: Boosting Scale Generalization for Object Detectors Without Inference Overhead

    by Wangzhong Li, Biao Li, Xuefeng Rao, Xiangmeng Pan, Yujie Zhao

    In object detection tasks, scale variation is one of the major challenges affecting detection accuracy. Existing methods typically enhance the scale generalization of models through strategies through image pyramids, feature pyramid networks, or multi-scale training. However, these approaches either increase inference computational costs or have not previously established explicit associations between different scale inputs during training. We propose Dual-Scale Parallel Training (DSPT) as a novel training strategy for object detection models. This method introduces high-and-low-resolution homologous image branches for associative learning during training. By fusing the losses from both branches, DSPT enables shared-parameter, same-iteration cross-scale joint optimization strategy, thereby enhancing the model’s multi-scale perception and generalization ability without altering the network structure or inference efficiency. Experimental results demonstrate that DSPT achieves stable performance improvements across several popular datasets (MS COCO, PASCAL VOC, VisDrone, etc.) and two model series (YOLO and RTDETR), with the mAP@50 metric improving by up to 4%. In scale-aware evaluation experiments, DSPT outperforms conventional and multi-scale training methods in detection accuracy and generalization, and provides positive gains across all object sizes. This strategy is simple, general, and scalable, providing a new approach to training object detection models with promising practical applications.