李楚琳,时启龙,杨朝阳

• 1:湖北汽车工业学院汽车工程学院

摘要(Abstract)：

针对实时检测Transformer(real-time detection transformer, RT-DETR)在模型参数量大、目标尺度变化显著时，检测精度不高，易产生误检、漏检等问题，本文提出基于RT-DETR的轻量化交通目标检测改进算法DTDETR(dynamic inception&token statistics detection transformer)。首先，使用DIMConv改进主干网络残差块中的3×3卷积，采用带状和块状卷积核相结合的方式提取特征，再通过动态权重生成机制将特征信息加权融合，从而降低计算成本和提高交通目标的检测精度。其次，在混合编码器部分用TSSA模块替换AIFI模块，进一步提升模型检测速度。在SODA10M数据集上的实验结果表明，DT-DETR模型在实现轻量化的同时，检测性能和推理速度显著提升；相比原RT-DETR模型，DT-DETR模型参数量减少34.2%，计算量降低23.3%,F1分数提高1.99%，推理速度提升130.9%，为自动驾驶目标检测提供了更高效、准确且轻量化的解决方案。

关键词(KeyWords)： 目标检测;RT-DETR;轻量化;DIMConv;TSSA

Abstract:

Keywords:

基金项目(Foundation): 江苏省自然科学基金项目（BK20230545）资助

作者(Author): 李楚琳,时启龙,杨朝阳

DOI: 10.16375/j.cnki.cn45-1395/t.2026.03.010

参考文献(References)：

• [1]REDMON J,DIVVALA S,GIRSHICK R,et al. You only look once:unified , real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR). June 27-30,2016. Las Vegas,NV,USA. IEEE,2016:779-788.
• [2]REDMON J,FARHADI A.YOLO9000:better,faster,stronger[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR). July21-26,2017. Honolulu,HI,USA. IEEE,2017:7263-7271.
• [3]BOCHKOVSKIY A,WANG C Y,LIAO H. YOLOv4:optimal speed and accuracy of object detection[J]. arXiv preprint arXiv:2004. 10934,2020.
• [4]JIA X,TONG Y,QIAO H,et al. Fast and accurate object detector for autonomous driving based on improved YOLOv5[J]. Scientific Reports,2023,13(1):9711.
• [5]周爱玲，谭光兴.基于YOLOv5s的交通信号灯检测算法[J].广西科技大学学报，2023,34(4):69-76.
• [6]张成涛，李习刊，徐伟航，等.基于改进YOLOv8的复杂路况下的目标识别[J].广西科技大学学报，2025,36(3):85-91,122.
• [7]REN S Q,HE K M,GIRSHICK R,et al. Faster R-CNN:towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(6):1137-1149.
• [8]ZHOU Y,WEN S J,WANG D L,et al. Object detection in autonomous driving scenarios based on an improved faster-RCNN[J]. Applied Sciences,2021,11(24):11630.
• [9]VASWANI A,SHAZEER N,PARMAR N,et al. Attention is all you need[C]//Advances in Neural Information Processing Systems. Long Beach , CA , USA. Curran Associates,Inc.,2017:5998-6008.
• [10]ZHAO Y,LV W,XU S,et al. DETRs beat YOLOs on real-time object detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle,WA,USA. IEEE,2024:1-10.
• [11]沈烨，王泉，吴昊. FSSD-DETR自动驾驶场景实时目标检测算法[J].电子测量技术，2024,47(14):88-95.
• [12]DING X H,ZHANG X Y,MA N N,et al. RepVGG:making VGG-style ConvNets great again[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR). June 20-25,2021.Nashville,TN,USA.IEEE,2021:13733-13742.
• [13]SZEGEDY C,LIU W,JIA Y Q,et al. Going deeper with convolutions[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR). June 7-12,2015. Boston,MA,USA. IEEE,2015:1-9.
• [14]SZEGEDY C , VANHOUCKE V , IOFFE S , et al.Rethinking the inception architecture for computer vision[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition(CVPR). June 27-30,2016.Las Vegas,NV,USA. IEEE,2016:2818-2826.
• [15]SZEGEDY C , IOFFE S , VANHOUCKE V , et al.Inception-v4,inception-ResNet and the impact of residual connections on learning[C]//Proceedings of the AAAI Conference on Artificial Intelligence,San Francisco,CA,USA:AAAI,2017,31(1):1-10. https：//arxiv.org/pdf/1602.07261.
• [16]WU Z Y,DING T J,LU Y F,et al. Token statistics transformer:linear-time attention via variational rate reduction[J]. arXiv preprint arXiv:2412. 17810,2024.
• [17]HAN J H,LIANG X W,XU H,et al. SODA10M:towards large-scale object detection benchmark for autonomous driving[J]. arXiv preprint arXiv:2106. 11118,2021.

文章评论(Comment)：

序号(No.) 时间(Time) 反馈人(User) 邮箱(Email) 标题(Title) 内容(Content)

反馈人(User) 邮箱地址(Email) 反馈标题(Title) 反馈内容(Content)