远距离情形下的改进YOLOv8行人检测算法

doi:10.19678/j.issn.1000-3428.0068897

摘要/Abstract

摘要：

智慧社区场景下的行人检测需要精准识别行人以应对各类情况的发生, 然而面对遮挡和远距离行人的情景, 现有检测器会出现漏检、误检以及模型过大不易部署的问题。针对以上问题, 提出基于YOLOv8的行人检测算法ME-YOLO。设计一种高效特征提取模块(EM), 使得网络更好地学习行人特征和捕捉行人特点, 在减少网络参数量的同时提高检测精度。设计一个重构的检测头模块, 重新整合后的检测层增强了网络对小目标的识别能力, 有效检测小目标行人。引入双向特征金字塔网络来设计新的颈部网络, 即双向扩张残差-特征金字塔网络(BDR-FPN), 利用扩张残差模块和附权注意力机制来扩展感受野及有所侧重地学习行人特征, 缓解网络对遮挡行人不敏感问题。实验结果表明, 在CityPersons数据集上进行训练和验证, 相比原算法YOLOv8, ME-YOLO算法的AP₅₀提高了5.6百分点, 模型参数量减少了41%, 模型大小压缩了40%, 在TinyPerson数据集上验证算法的有效性和泛化性, AP₅₀提高了4.1百分点, AP_50∶95提高了1.7百分点。该算法在大幅度减少模型参数和大小的同时, 有效提高了检测精度, 在智慧社区场景中有较好的应用价值。

关键词: 行人检测, 智慧社区, 小目标行人, 特征金字塔网络, YOLOv8算法

Abstract:

Pedestrian detection in intelligent community scenarios needs to accurately recognize pedestrians to address various situations. However, for persons who are occluded or at long distances, existing detectors exhibit problems such as missed detection, detection error, and large models. To address these problems, this paper proposes a pedestrian detection algorithm, Multiscale Efficient-YOLO (ME-YOLO), based on YOLOv8. An efficient feature Extraction Module (EM) is designed to improve network learning and capture pedestrian features, which reduces the number of network parameters and improves detection accuracy. The reconstructed detection head module reintegrates the detection layer to enhance the network's ability to recognize small targets and effectively detect small target pedestrians. A Bidirectional Feature Pyramid Network (BiFPN) is introduced to design a new neck network, namely the Bidirectional Dilated Residual-Feature Pyramid Network (BDR-FPN), and the expanded residual module and weighted attention mechanism expand the receptive field and learn pedestrian features with emphasis, thereby alleviating the problem of network insensitivity to occluded pedestrians. Compared with the original YOLOv8 algorithm, ME-YOLO increases the AP₅₀ by 5.6 percentage points, reduces the number of model parameters by 41%, and compresses the model size by 40% after training and verification based on the CityPersons dataset. ME-YOLO also increases the AP₅₀ by 4.1 percentage points and AP_50∶95 by 1.7 percentage points on the TinyPerson dataset. Moreover, the algorithm significantly reduces the number of model parameters and model size and effectively improves detection accuracy. This method has a considerable application value in intelligent community scenarios.

Key words: pedestrian detection, intelligent community, small target pedestrian, Feature Pyramid Network (FPN), YOLOv8 algorithm

汤静雯, 赖惠成, 王同官. 远距离情形下的改进YOLOv8行人检测算法[J]. 计算机工程, 2025, 51(4): 303-313.

TANG Jingwen, LAI Huicheng, WANG Tongguan. Improved YOLOv8 Pedestrian Detection Algorithm for Long-Distance Situations[J]. Computer Engineering, 2025, 51(4): 303-313.

https://www.ecice06.com/CN/Y2025/V51/I4/303

图/表 20

图1 YOLOv8算法网络结构

Fig.1 Network structure YOLOv8 algorithm

图2 ME-YOLO算法网络结构

Fig.2 Network structure ME-YOLO algorithm

图3 GhostConv原理图

Fig.3 GhostConv schematic diagram

图4 EC原理图

Fig.4 EC schematic diagram

图5 EM模块结构

Fig.5 EM module structure

图6 重构检测头结构

Fig.6 Reconstructed detection head structure

图7 改进的特征金字塔结构

Fig.7 Improved feature pyramid structure

图8 扩张残差模块结构

Fig.8 Dilation-wise residual module structure

图9 改进颈部网络BDR-FPN

Fig.9 Improved neck network BDR-FPN

图10 YOLOv8s和ME-YOLO的平均精度比较

Fig.10 Average accuracy comparison between YOLOv8s and ME-YOLO

图11 不同检测层的检测效果

Fig.11 The detection effect of different detection layers

图12 普通场景下YOLOv8s和ME-YOLO的检测效果比较

Fig.12 Comparison of detection effect between YOLOv8s and ME-YOLO in common scenes

图13 暗光条件下YOLOv8s和ME-YOLO的检测效果比较

Fig.13 Comparison of detection effect of YOLOv8s and ME-YOLO under low light condition

图14 YOLOv8s和ME-YOLO的小目标行人检测效果比较

Fig.14 Comparison of small-target pedestrian detection effect of YOLOv8s and ME-YOLO

参考文献 25

1	ZHANG W C , FU C , XIE H Y , et al. Global context aware RCNN for object detection. Neural Computing and Applications, 2021, 33 (18): 11627- 11639. doi: 10.1007/s00521-021-05867-1
2	ARORA N , KUMAR Y , KARKRA R , et al. Automatic vehicle detection system in different environment conditions using fast R-CNN. Multimedia Tools and Applications, 2022, 81 (13): 18715- 18735. doi: 10.1007/s11042-022-12347-8
3	LI X M , XIE Z J , DENG X , et al. Traffic sign detection based on improved faster R-CNN for autonomous driving. The Journal of Supercomputing, 2022, 78 (6): 7982- 8002. doi: 10.1007/s11227-021-04230-4
4	GAWANDE U , HAJARI K , GOLHAR Y . SIRA: Scale illumination rotation affine invariant mask R-CNN for pedestrian detection. Applied Intelligence, 2022, 52 (9): 10398- 10416. doi: 10.1007/s10489-021-03073-z
5	LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection[C]//Proceedings of IEEE International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2017: 2980-2988.
6	LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot MultiBox detector[C]//Proceedings of European Conference on Computer Vision. Berlin, Germany: Springer, 2016: 21-37.
7	LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot MultiBox detector[C]//Proceedings of European Conference on Computer Vision. Berlin, Germany: Springer, 2016: 21-37.
8	裴伟, 许晏铭, 朱永英, 等. 改进的SSD航拍目标检测方法. 软件学报, 2019, 30 (3): 738- 758.
	PEI W , XU Y M , ZHU Y Y , et al. The target detection method of aerial photography images with improved SSD. Journal of Software, 2019, 30 (3): 738- 758.
9	DONG C , LUO X S . Research on a pedestrian detection algorithm based on improved SSD network. Journal of Physics: Conference Series, 2021, 1802 (3): 032073. doi: 10.1088/1742-6596/1802/3/032073
10	高宗, 李少波, 陈济楠, 等. 基于YOLO网络的行人检测方法. 计算机工程, 2018, 44 (5): 215-219, 226. doi: 10.19678/j.issn.1000-3428.0046885
	GAO Z , LI S B , CHEN J N , et al. Pedestrian detection method based on YOLO network. Computer Engineering, 2018, 44 (5): 215-219, 226. doi: 10.19678/j.issn.1000-3428.0046885
11	徐守坤, 邱亮, 李宁, 等. 基于HOG-CSLBP及YOLOv2的行人检测. 计算机工程与设计, 2019, 40 (10): 2964- 2968.
	XU S K , QIU L , LI N , et al. Pedestrian detection based on HOG-CSLBP and YOLOv2. Computer Engineering and Design, 2019, 40 (10): 2964- 2968.
12	魏润辰, 何宁, 尹晓杰. YOLO-Person: 道路区域行人检测. 计算机工程与应用, 2020, 56 (19): 197- 204.
	WEI R C , HE N , YIN X J . YOLO-person: pedestrian detection in road areas. Computer Engineering and Applications, 2020, 56 (19): 197- 204.
13	陈一潇, 阿里甫·库尔班, 林文龙, 等. 面向拥挤行人检测的CA-YOLOv5. 计算机工程与应用, 2022, 58 (9): 238- 245.
	CHEN Y X , Alifu·Kuerban , LIN W L , et al. CA-YOLOv5 for crowded pedestrian detection. Computer Engineering and Applications, 2022, 58 (9): 238- 245.
14	HOU Q B, ZHOU D Q, FENG J S. Coordinate attention for efficient mobile network design[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2021: 13713-13722.
15	LIN T Y, DOLLAR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2017: 1063-6919.
16	TANG F , YANG F , TIAN X Q . Long-distance person detection based on YOLOv7. Electronics, 2023, 12 (6): 1502. doi: 10.3390/electronics12061502
17	Ultralytics. UltralyticsYOLOv8[EB/OL]. [2023-10-20]. https://github.com/ultralytics/ultralytics.
18	TAN M X, PANG R M, LE Q V. EfficientDet: scalable and efficient object detection[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2020: 10781-10790.
19	SONG G, LIU Y, WANG X. Revisiting the sibling head in object detector[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2020: 257-268.
20	HAN K, WANG Y H, TIAN Q, et al. GhostNet: more features from cheap operations[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2020: 1580-1589.
21	HOWARD A, SANDLER M, CHU G, et al. Searching for MobileNetV3[EB/OL]. [2023-10-20]. https://arxiv.org/pdf/1905.02244v3.
22	HOWARD A G, ZHU M L, CHEN B, et al. MobileNets: efficient convolutional neural networks for mobile vision applications[EB/OL]. [2023-10-20]. https://arxiv.org/pdf/1704.04861.
23	MA N N, ZHANG X Y, ZHENG H T, et al. ShuffleNet V2: practical guidelines for efficient CNN architecture design[C]//Proceedings of European Conference on Computer Vision. Berlin, Germany: Springer, 2018: 122-138.
24	CORDTS M, OMRAN M, RAMOS S, et al. The cityscapes dataset for semantic urban scene understanding[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2016: 3213-3223.
25	YU X H, GONG Y Q, JIANG N, et al. Scale match for tiny person detection[C]//Proceedings of IEEE Winter Conference on Applications of Computer Vision. Washington D. C., USA: IEEE Press, 2020: 1257-1265.

[1]	胡倩, 皮建勇, 胡伟超, 黄昆, 王娟敏. 基于改进YOLOv5的密集行人检测算法[J]. 计算机工程, 2025, 51(3): 216-228.
[2]	贵向泉, 刘世清, 李立, 秦庆松, 李唐艳. 基于改进YOLOv8的景区行人检测算法[J]. 计算机工程, 2024, 50(7): 342-351.
[3]	赵继达, 甄国涌, 储成群. 基于YOLOv8的无人机图像目标检测算法[J]. 计算机工程, 2024, 50(4): 113-120.
[4]	徐芳芯, 樊嵘, 马小陆. 面向拥挤行人检测的改进YOLOv7算法[J]. 计算机工程, 2024, 50(3): 250-258.
[5]	龙玉江, 卫薇, 舒彧, 张正刚, 王道累, 李峰. 基于自适应关键点的破损旋转绝缘子检测方法[J]. 计算机工程, 2023, 49(9): 272-278.
[6]	王程, 刘元盛, 刘圣杰. 基于改进YOLOv4的小目标行人检测算法[J]. 计算机工程, 2023, 49(2): 296-302,313.
[7]	邹慧海, 侯进. 改进SSD算法的道路小目标检测研究[J]. 计算机工程, 2022, 48(5): 281-288.
[8]	候瑞环, 杨喜旺, 王智超, 高佳鑫. 一种基于YOLOv4-TIA的林业害虫实时检测方法[J]. 计算机工程, 2022, 48(4): 255-261.
[9]	汪常建, 丁勇, 卢盼成. 融合改进FPN与关联网络的Faster R-CNN目标检测[J]. 计算机工程, 2022, 48(2): 173-179.
[10]	施政, 毛力, 孙俊. 基于YOLO的多模态加权融合行人检测算法[J]. 计算机工程, 2021, 47(8): 234-242.
[11]	音松, 陈雪云, 贝学宇. 改进Mask RCNN算法及其在行人实例分割中的应用[J]. 计算机工程, 2021, 47(6): 271-276,283.
[12]	文韬, 周稻祥, 李明. Mask R-CNN中特征不平衡问题的全局信息融合方法[J]. 计算机工程, 2021, 47(3): 256-260,268.
[13]	李鑫然, 李书琴, 刘斌. 基于改进Faster R_CNN的苹果叶片病害检测模型[J]. 计算机工程, 2021, 47(11): 298-304.
[14]	姜建勇, 吴云, 龙慧云, 黄自萌, 蓝林. 基于CenterNet的实时行人检测模型[J]. 计算机工程, 2021, 47(10): 276-282.
[15]	陈泽, 叶学义, 钱丁炜, 魏阳洋. 基于改进Faster R-CNN的小尺度行人检测[J]. 计算机工程, 2020, 46(9): 226-232,241.

选择文件类型/文献管理软件名称

选择包含的内容