| 38 | 0 | 73 |
| 下载次数 | 被引频次 | 阅读次数 |
为改善复杂道路场景中小目标漏检、误检问题，提出了一种基于改进YOLOv8的复杂道路场景目标检测算法。首先，在特征提取网络中引入RFCAConv模块，以提升模型对关键特征信息的抓取能力；其次，为丰富深层次网络中的上下文信息，在聚合网络中构建了上下文感知模块CAM；最后，将原本的损失函数CIoU替换为Inner-CIoU，以增强模型的泛化能力。实验结果表明：在New＿BDD公开数据集上，本算法模型检测精度mAP可达77.6%，推理速度FPS达98帧/s，优于其他主流算法。同时，在图像检测结果的可视化实验中，该算法模型在面对多种复杂道路场景的图像实例中有效改善了原有模型的漏检和误检等情况。
Abstract：To address the issues of missed and false detection of small objects in complex road scenes,this paper proposes a complex road scene object detection algorithm based on YOLOv8.Firstly,a RFCAConv module is introduced into the feature extraction network to enhance the model's ability to capture key feature information;secondly,to enrich the contextual information in deeper networks,a Context-Aware Module(CAM)is constructed in the aggregation network;finally,the original CIoU loss function is replaced with Inner-CIoU to improve the model's generalization ability.Experimental results show that on the New＿BDD public datasets,the proposed algorithm achieves a detection precision(mAP)of 77.6%and an inference speed(FPS)of 98 frames per second,outperforming other mainstream algorithms.Additionally,in the visualization experiments of image detection results,the proposed model effectively reduces missed and false detection in various complex road scene image instances compared to the original model.
[1]寻永恒，王旭成，杨兴平．智能网联汽车环境感知技术研究[J]．汽车知识，2025,25(8):31-33.
[2]REN S,HE K,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.
[3]HE K,GKIOXARI G,DOLLÁR P,et al.Mask R-CNN[C]//2017IEEE international conference on computer vision.Venice:IEEE,2017:2980-2988.
[4]LIU W,ANGUELOV D,ERHAN D,et al.Ssd:single shot multibox detector[C]//LEIBE B,MATAS J,WELING M(eds).Computer Vision-ECCV 2016.Cham:Springer,2016:21-37.
[5]REDMON J,DIVVALA S,GIRSHICK R,et al.You only look once:unified,real-time object detection[C]//2016IEEE conference on computer vision and pattern recognition.Las Vegas:IEEE,2016:779-788.
[6]REDMON J,FARHADI A.YOLO9000:better,faster,stronger[C]//2017IEEE conference on computer vision and pattern recognition.Honolulu:IEEE,2017:6517-6525.
[7]REDMON J,FARHADI A.Yolov3:an incremental improvement[PP/OL].V1.(2018-04-08)[2025-07-23].https：//arxiv.org/abs/1804.02767.
[8]薛阳，徐笑，卢秋红，等．针对光照干扰的TSA-YOLO车辆检测算法[J]．计算机工程与设计，2025,46(7):2021-2027.
[9]霍华，邢晓钰．复杂场景下的轻量化行人检测算法研究[J/OL]．计算机应用与软件（2025-07-04）[2025-08-02].https：//link.cnki.net/urlid/31.1260.tp.20250703.1757.002.
[10]孙海明，付世超．基于改进YOLOv4-Tiny的交通标志图像识别算法研究[J]．计算机应用与软件，2025,42(5):164-170.
[11]SONG G,LIU Y,WANG X.Revisiting the sibling head in object detector[C]//2020IEEE/CVF conference on computer vision and pattern recognition.Seattle:IEEE,2020:11560-11569.
[12]ZHANG X,LIU C,YANG D,et al.RFAConv:innovating spatial attention and standard convolutional operation[PP/OL].V4.(2023-04-18)[2025-07-23].https：//arxiv.org/abs/2304.03198.
[13]HE K,ZHANG X,REN S,et al.Deep residual learning for image recognition[C].20116IEEE conference on computer vision and pattern recognition.Las Vegas:IEEE,2016:770-778.
[14]ZHANG H,XU C,ZHANG S.Inner-IoU:more effective intersection over union loss with auxiliary bounding box[PP/OL].V4.(2023-11-14)[2025-07-23].https：//arxiv.org/abs/2311.02877.
[15]YU F,CHEN H F,WANG X,et al.BDD100K:a diverse driving dataset for heterogeneous multitask learning[C]//2020IEEE/CVF Conference on computer vision and pattern recognition,Seattle:IEEE,2020:2633-2642．
基本信息:
DOI：10.13486/j.issn.2097-4973.2026.02.009
中图分类号:U495;TP391.41
引用信息:
[1]董凤,张鑫,孔林,等.基于改进YOLOv8的复杂道路场景目标检测算法[J].山东航空学院学报,2026,43(02):73-81.DOI:10.13486/j.issn.2097-4973.2026.02.009.
基金信息:
信阳学院校级科研项目（2024-XJLYB-009）
2026-04-15
2026-04-15