College of Railway Transportation,Hunan University of Technology,Zhuzhou 412007,China
Show less
文章历史+
Received
Published
Revised
2025-02-26
2026-02-08
2025-10-31
Online Date
2026-09-21
PDF (25851K)
摘要
针对现有目标检测算法存在对交通标志中的小目标识别准确率低、检测不完全等问题,本文提出一种基于改进YOLO11n的交通标志检测算法。首先,将卷积和注意力融合模块(convolution and attention fusion module,CAFM)与YOLO11n主干部分结合,通过融合卷积操作和注意力机制,对图像的全局和局部特征进行有效建模,使小目标的特征表达更为充分;其次,引入轻量动态上采样算法DySample,使采样点集中于目标区域,减少特征信息的丢失,进而增强对小目标的感知能力;然后,使用选择性卷积(selective convolution,SCConv)与C3k2相结合,设计全新的C3k2_SC模块,提高检测精度,减少模型参数量;最后,添加一个小目标检测层,保留更多浅层细节信息,以增强深层和浅层语义信息的融合,从而改善了对小目标检测不完全的问题。实验结果证明,改进后的算法在TT100K数据集中,精确率、召回率、平均精度分别达到82.0%、77.5%、84.6%,与原模型YOLO11n相比较,提高了3.8、9.5、8.5个百分点,证明了改进的有效性。
Abstract
Aiming at the existing target detection algorithms,which suffer from problems such as low accuracy and incomplete detection of small targets in traffic signs,a traffic sign detection algorithm based on an improved YOLO11n was proposed. Initially,a convolution and attention fusion module (CAFM) was integrated into the backbone network of YOLO11n to comprehensively model global and local image features through the combined operation of convolutional layers and attention mechanisms,thereby improving small target feature representation. Subsequently,the lightweight dynamic up-sampling algorithm DySample was introduced to optimize feature sampling by concentrating sampling points on target regions,effectively mitigating feature information loss and enhancing small target perception. Then,a novel C3k2_SC module was designed by incorporating selective convolution (SCConv) into the C3k2 structure,which improved detection accuracy while reducing model parameter complexity. Finally,an additional small target detection layer was added to preserve shallow-level detailed information and strengthen the fusion of deep and shallow semantic features,thereby resolving incomplete detection of small targets. The experimental results demonstrate that precision,recall,and mean average precision rates of 82.0%,77.5%,and 84.6% are achieved by the improved algorithm on the TT100K dataset,representing improvements of 3.8,9.5,and 8.5 percentage points compared to the original YOLO11n model,proving the effectiveness of the improvement.
MouJiayu, NanXinyuan. Improved real-time traffic sign small target detection algorithm based on YOLOv7-tiny[J]. Science Technology and Engineering, 2024, 24(30): 13072-13079.
CaoJiale, LiYali, SunHanqing, et al. A review of visual target detection techniques based on deep learning[J]. Chinese Journal of Image Graphics, 2022, 27(6): 1697-1722.
[6]
LiangT, BaoH, PanW, et al. Traffic sign detection via improved sparse R-CNN for autonomous vehicles[J]. Journal of Advanced Transportation, 2022, 9: 3825532.
[7]
JiangH, Learned-MillerE. Face detection with the faster R-CNN[C]// The Proceedings of 2017 12th IEEE International Conference on Automatic Face & Gesture Recognition. Washington, DC: IEEE, 2017: 650-657.
[8]
LiangZ, ShaoJ, ZhangD, et al. Traffic sign detection and recognition based on pyramidal convolutional networks[J]. Neural Computing and Applications, 2020, 32(11): 6533-6543.
[9]
ZhuZ, LiangD, ZhangS, et al. Traffic-sign detection and classification in the wild[C]// The Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Lasvegas: IEEE, 2016: 2110-2118.
SongJingjing, ZhangZhenli. Small target traffic sign detection algorithm based on YOLOv8n[J]. Science Technology and Engineering, 2024, 24(36): 15548-15557.
WuChenxi, YingBaosheng, XuXiaowei, et al. A small target detection algorithm for roads based on improved single-step multi-frame target detection[J]. Science Technology and Engineering, 2023, 23(5): 2051-2058.
[14]
RajendranS P, ShineL, PradeepR, et al. Real-time traffic sign recognition using YOLOv3 based detector[C]// The Proceedings of 2019 10th International Conference on Computing, Communication and Networking Technologies (ICCCNT). Kanpur: IEEE, 2019: 1-7.
PanGuixia, LaiHuicheng, WangTongguan, et al. A YOLOV5 neck refinement algorithm for small traffic sign detection[J]. Modern Electronics Technique, 2023, 46(14): 56-62.
ZhaoHuipeng, CaoJingsheng, PanDijing, et al. Improved YOLOv8 algorithm for small target detection of traffic signs[J]. Modern Electronics Technique, 2024, 47(20): 141-147.
[25]
HuB, ChenS. Edge-guided contextual attention fusion network for camouflaged object detection[C]// The Proceedings of 2024 3rd International Conference on Cyber Security, Artificial Intelligence and Digital Economy. Nanjing: Association for Computing Machinery, 2024: 108-112.
[26]
LiuW, LuH, FuH, et al. Learning to upsample by learning to sample[C]// The Proceedings of IEEE/CVF International Conference on Computer Vision. Paris: IEEE, 2023: 6027-6037.
[27]
YeS, DongB, XiongW, et al. A deep convolutional neural network for blind element error correction of spatial heterodyne spectrometer using line selective convolutional blocks[J]. Journal of Quantitative Spectroscopy and Radiative Transfer, 2025, 330: 109199.
MengFangfang, TianXiaozhuang, FangWei, et al. A high-altitude nut recognition algorithm based on improved YOLOv5[J]. Science Technology and Engineering, 2025, 25(1): 262-269.
NiuSiqi, MaRui, XuXiaolin, et al. MobileNetV2 maize seed variety recognition based on improved CBAM attention mechanism[J]. Journal of the Chinese Cereals and Oils Association, 2024, 39(3): 159-165.
[32]
WangJ, ChenK, XuR, et al. Carafe: content-aware reassembly of features[C]// The Proceedings of IEEE/CVF International Conference on Computer Vision. Seoul: IEEE, 2019: 3007-3016.
[33]
RahmanM M, MunirM, MarculescuR. Emcad: efficient multi-scale convolutional attention decoding for medical image segmentation[C]// The Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Austintexas: IEEE, 2024: 11769-11779.
ZhaoHong, FengYubo. Research on traffic sign detection based on CGS-Ghost YOLO[J]. Computer Engineering, 2023, 49(12): 194-204.
[39]
ZhangJ, ZouX, KuangL D, et al. CCTSDB 2021: a more comprehensive traffic sign detection benchmark[J]. Human-centric Computing and Information Sciences, 2022, 12: 23.