一种面向铁路道口嵌入式设备的轻量级目标检测算法
A lightweight object detection algorithm for railway crossing embedded devices
赵阳 1王彬 1楼向东 1赵士豪 2宋保业2
作者信息
- 1. 兖矿物流科技有限公司铁路分公司，山东邹城 273500
- 2. 山东科技大学电气与自动化工程学院，山东青岛 266590
- 折叠
摘要
为解决铁路道口嵌入式设备上目标检测算法部署困难的问题，提出一种轻量级算法--YOLO-ED。该算法通过引入新颖的C3-Faster模块、坐标注意力机制及Focal-EIoU损失函数对YOLOV5进行优化，旨在实现检测精度与计算效率的平衡。实验结果表明，与基线模型相比，YOLO-ED在保持检测精度基本不变的前提下，推理速度提升26.9%，模型参数量和计算量分别显著降低27.3%和33.3%。最终在Jetson Nano B01平台上成功部署，进一步验证了该算法在资源受限场景下的高效性与工程应用价值。
Abstract
To address the difficulty in deploying object detection algorithms on embedded devices at railway crossings， a lightweight algorithm， namely YOLO-ED， was proposed.The YOLOV5 model was optimized through the introduction of a novel C3-Faster module， a coordinate attention mechanism， and Focal-EIoU loss function， aiming at achieving a balance between detection accuracy and computational efficiency. The experimental results demonstrate that， in comparison with the baseline model， the inference speed of YOLO-ED is increased by 26.9% while maintaining a comparable level of detection accuracy， with the parameter count and computational load significantly reduced by 27.3% and 33.3%， respectively. The successful deployment of the algorithm on the Jetson Nano B01 platform further validates its high efficiency and practical engineering value in resource-constrained scenarios.关键词
铁路道口/嵌入式系统/轻量化目标检测算法/YOLOKey words
railway level crossing/embedded systems/lightweight object detection algorithm/YOLO引用本文复制引用
赵阳,王彬,楼向东,赵士豪,宋保业.一种面向铁路道口嵌入式设备的轻量级目标检测算法[EB/OL].(2025-11-28)[2026-04-02].http://www.paper.edu.cn/releasepaper/content/202511-39.学科分类
电子技术应用
评论