PATD-YOLO：基于YOLOv11的道路障碍目标检测算法

doi:10.3778/j.issn.1002-8331.2504-0038

摘要/Abstract

摘要： 道路障碍检测对车辆驾驶操控的稳定性与行驶的安全性极为重要，针对路面复杂纹理和异物多样性导致的误检、漏检及小目标物体检测精度不足问题，提出基于YOLOv11的改进道路障碍目标检测模型（PATD-YOLO）。设计融合双主干网络，其中MCPStem（multi-convolution pooling）模块共享浅层特征，HGNetV2作为第二主干引入动态卷积增强参数表达能力，特征动态校准融合（feature dynamic aligning merge）模块消除多源特征空间差异，实现高效特征融合。设计小目标物体增强特征金字塔，增加P2层高分辨率特征图，采用SPDConv切片操作保留小目标信息，设计GLAFI（global-local aggregate feature information）模块进行特征整合模块融合浅层细粒度与深层语义特征。引入可重参数化卷积（reparameterizable convolution）设计了重参数轻量化检测头（reparameterized lightweight detection head），提高参数利用率。实验结果表明，在参数量与计算量基本不变的情况下，改进算法在自建数据集与RDD2022公开数据集上的检测精度mAP@0.5相较于基准算法分别提升3.0和4.3个百分点，均超越了主流算法，并且检测速度达到131.6 FPS，能够满足车辆检测的实时性要求。

关键词: 道路障碍检测, YOLOv11, 特征动态融合, 可重参数化

Abstract: Road obstacle detection is crucial for ensuring driving stability and traffic safety. To address the issues of false detection, missed detection, and insufficient accuracy in small object detection caused by complex road textures and diverse obstacles, this paper proposes an improved road obstacle detection model based on YOLOv11, named PATD-YOLO. A dual-backbone network is designed, in which the multi-convolution pooling stem?(MCPStem) module shares shallow features, while HGNetV2 is introduced as the second backbone to enhance parameter representation capability through dynamic convolution. The feature dynamic aligning merge (FDAM) module eliminates spatial discrepancies in multi-source features, enabling efficient feature fusion. A small-object-enhanced feature pyramid is constructed by incorporating high-resolution P2-layer feature maps. SPDConv operations are employed to preserve small object information, and the global-local aggregate feature information (GLAFI) module is designed to integrate fine-grained shallow features with deep semantic features. A reparameterized lightweight detection head is introduced using reparameterizable convolution to improve parameter efficiency. Experimental results demonstrate that, with nearly identical parameters and computational costs, the proposed algorithm achieves 3.0 and 4.3 percentage points improvements in mAP@0.5 on a self-built dataset and the public RDD2022 dataset, respectively, outperforming mainstream algorithms. Moreover, the detection speed reaches 131.6 FPS, meeting real-time requirements for vehicle detection.

Key words: road obstacle detection, YOLOv11, feature dynamic fusion, reparameterizable

纪文杰, 宋廷伦, 容小洛, 周迈. PATD-YOLO：基于YOLOv11的道路障碍目标检测算法[J]. 计算机工程与应用, 2025, 61(21): 129-143.

JI Wenjie, SONG Tinglun, RONG Xiaoluo, ZHOU Mai. PATD-YOLO：Road Obstacle Object Detection Algorithm Based on YOLOv11[J]. Computer Engineering and Applications, 2025, 61(21): 129-143.

参考文献

[1] 王佳荣. 面向自动驾驶的多传感器三维环境感知系统关键技术研究[D]. 长春: 中国科学院大学 (中国科学院长春光学精密机械与物理研究所), 2020.
WANG J R. Research on key technologies of multi-sensor 3D environment perception systems for autonomous driving[D]. Changchun: Chinese Academy of Sciences (Changchun Institute of Optics, Fine Mechanics and Physics), 2020.
[2] ZHAO T, HE J X, LV J C, et al. A comprehensive implementation of road surface classification for vehicle driving assistance: dataset, models, and deployment[J]. IEEE Transactions on Intelligent Transportation Systems, 2023, 24(8): 8361-8370.
[3] 孟波, 史旭华, 张彬. 基于双分支卷积和深度插值的点云表面重建[J]. 计算机工程, 2025, 51(7): 119-126.
MENG B, SHI X H, ZHANG B. Point cloud surface reconstruction based on dual-branch convolution and deep interpolation[J]. Computer Engineering, 2025, 51(7): 119-126.
[4] LI Z J, KOLMANOVSKY I, ATKINS E, et al. H∞ filtering for cloud-aided semi-active suspension with delayed road information[J]. IFAC-PapersOnLine, 2015, 48(12): 275-280.
[5] TSAI Y J, CHATTERJEE A. Pothole detection and classification using 3D technology and watershed method[J]. Journal of Computing in Civil Engineering, 2018, 32(2): 04017078.
[6] HOU C J, XIE Y Q, ZHANG Z Z. An improved convolutional neural network based indoor localization by using Jenks natural breaks algorithm[J]. China Communications, 2022, 19(4): 291-301.
[7] HAN K, WANG Y H, CHEN H T, et al. A survey on vision transformer[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(1): 87-110.
[8] CARION N, MASSA F, SYNNAEVE G, et al. End-to-end object detection with transformers[C]//Proceedings of the European Conference on Computer Vision. Cham: Springer, 2020: 213-229.
[9] ZHAO Y A, LV W Y, XU S L, et al. DETRs beat YOLOs on real-time object detection[C]//Proceedings of the 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2024: 16965-16974.
[10] REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2016: 779-788.
[11] LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot MultiBox detector[C]//Proceedings of the 14th European Conference on Computer Vision. Cham: Springer, 2016: 21-37.
[12] 梁天添, 杨淞淇, 钱振明. 基于改进YOLOv8s的恶劣天气车辆行人检测方法[J]. 电子测量技术, 2024, 47(9): 112-119.
LIANG T T, YANG S Q, QIAN Z M. Improved YOLOv8s method for vehicle and pedestrian detection in adverse weather[J]. Electronic Measurement Technology, 2024, 47(9): 112-119.
[13] 陈科圻, 朱志亮, 邓小明, 等. 多尺度目标检测的深度学习研究综述[J]. 软件学报, 2021, 32(4): 1201-1227.
CHEN K Q, ZHU Z L, DENG X M, et al. Deep learning for multi-scale object detection: a survey[J]. Journal of Software, 2021, 32(4): 1201-1227.
[14] ARYA D, MAEDA H, GHOSH S K, et al. RDD2022: a multi-national image dataset for automatic road damage detection[J]. Geoscience Data Journal, 2024, 11(4): 846-862.
[15] GOU C, PENG B, LI T R, et al. Pavement crack detection based on the improved faster-RCNN[C]//Proceedings of the 2019 IEEE 14th International Conference on Intelligent Systems and Knowledge Engineering. Piscataway: IEEE, 2019: 962-967.
[16] LI S B, HUANG Y J. Damage detection algorithm based on faster-RCNN[C]//Proceedings of the 2023 5th International Conference on Electronics and Communication, Network and Computer Technology. Piscataway: IEEE, 2023: 177-180.
[17] 马荣贵, 黄训燕, 董世浩. 基于改进YOLOv8的实时坑槽检测算法研究[J/OL]. 计算机工程, 2024: 1-10(2024-08-13)[2025-03-20]. https://link.cnki.net/doi/10.19678/j.issn.1000-3428.0069671.
MA R G, HUANG X Y, DONG S H. Research on real-time pothole detection algorithm based on improved YOLOv8[J/OL]. Computer Engineering, 2024: 1-10(2024-08-13)[2025-03-20]. https://link.cnki.net/doi/10.19678/j.issn.1000-3428.0069671.
[18] WANG C Y, YEH I H, LIAO H Y M. YOLOv9: learning what you want toLearn using programmable gradient information[C]//Proceedings of the European Conference on Computer Vision. Cham: Springer, 2025: 1-21.
[19] 刘泽, 宋廷伦, 石先让, 等. 基于改进RT-DETR的路面异常检测技术研究[J/OL]. 计算机工程, 2024: 1-14(2024-11-21)[2025-03-20]. https://link.cnki.net/doi/10.19678/j.issn.1000-3428.00EC0070182.
LIU Z, SONG T L, SHI X R, et al. Research on pavement anomaly detection technology based on improved RT-DETR[J/OL]. Computer Engineering, 2024: 1-14(2024-11-21)[2025-03-20]. https://link.cnki.net/doi/10.19678/j.issn.1000-3428.00EC0070182.
[20] 辛世澳, 葛海波, 袁昊, 等. 改进YOLOv7的轻量化水下目标检测算法[J]. 计算机工程与应用, 2024, 60(3): 88-99.
XIN S A, GE H B, YUAN H, et al. Improved lightweight underwater target detection algorithm of YOLOv7[J]. Computer Engineering and Applications, 2024, 60(3): 88-99.
[21] VORUGUNTI C S, PULABAIGARI V, GORTHI R K S S, et al. OSVFuseNet: online signature verification by feature fusion and depth-wise separable convolution based deep learning[J]. Neurocomputing, 2020, 409: 157-172.
[22] CHEN Y P, DAI X Y, LIU M C, et al. Dynamic convolution: attention over convolution kernels[C]//Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2020: 11027-11036.
[23] KOLAHI S G, CHAHARSOOGHI S K, KHATIBI T, et al. MSA2Net: multi-scale adaptive attention-guided network for medical image segmentation[J]. arXiv:2407.21640,2024.
[24] SUNKARA R, LUO T. No more strided convolutions or pooling: a new CNN building block for low-resolution images and small objects[C]//Proceeding of the Joint European Conference on Machine Learning and Knowledge Discovery in Databases. Cham: Springer, 2023: 443-459.
[25] CUI Y N, REN W Q, KNOLL A. Omni-kernel network for image restoration[C]//Proceedings of the AAAI Conference on Artificial Intelligence, 2024: 1426-1434.
[26] 丁桐. 基于深度学习的高分辨率遥感影像道路提取算法研究[D]. 哈尔滨: 黑龙江大学, 2024.
DING T. Research on high resolution remote sensing image road extraction algorithm based on deep learning[D]. Harbin: Heilongjiang University, 2024.
[27] TIAN Z, SHEN C H, CHEN H, et al. FCOS: fully convolutional one-stage object detection[C]//Proceedings of the 2019 IEEE/CVF International Conference on Computer Vision. Piscataway: IEEE, 2019: 9626-9635.
[28] DING X H, ZHANG X Y, MA N N, et al. RepVGG: making VGG-style ConvNets great again[C]//Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2021: 13728-13737.
[29] SAI GANESH NAIK M B, NIRMALRANI V. Detecting potholes using image processing techniques and real-world footage[C]//Proceeding of CISC 2020,Cognitive Informatics and Soft Computing. Singapore: Springer, 2021: 893-902.
[30] 吴明杰, 云利军, 陈载清, 等. 改进YOLOv5s的无人机视角下小目标检测算法[J]. 计算机工程与应用, 2024, 60(2): 191-199.
WU M J, YUN L J, CHEN Z Q, et al. Improved YOLOv5s small object detection algorithm in UAV view[J]. Computer Engineering and Applications, 2024, 60(2): 191-199.
[31] ZHAO K, PENG S F, LI Y J, et al. A lightweight Xray-YOLO-Mamba model for prohibited item detection in X-ray images using selective state space models[J]. Scientific Reports, 2025, 15: 13171.
[32] TAN M, LE Q. Efficientnet: rethinking model scaling for convolutional neural networks[C]//Proceedings of the International Conference on Machine Learning, 2019: 6105-6114.
[33] QIN D F, LEICHNER C, DELAKIS M, et al. MobileNetV4: universal models fortheMobile ecosystem[C]//Proceedings of the European Conference on Computer Vision. Cham: Springer, 2025: 78-96.
[34] FENG Y F, HUANG J G, DU S Y, et al. Hyper-YOLO: when visual object detection meets hypergraph computation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025, 47(4): 2388-2401.