改进的YOLOv8n轻量化景区行人检测方法研究

doi:10.3778/j.issn.1002-8331.2407-0402

摘要/Abstract

摘要： 针对景区人流量大、人员密集，而现有目标检测算法对于遮挡目标和小目标检测效率低且模型参数量大等问题，提出基于YOLOv8n的轻量化景区行人检测算法SSC-YOLOv8n。提出空间和通道重建注意力卷积SCC2fEMA模块，以显著减少模型参数量，从而提升模型的检测速度。采用精细的slim-neck范式，通过GSConv和V0V-GSCSP模块，在有效降低模型参数量的同时，提升模型的学习能力。提出坐标注意力动态解耦头，以显著增强模型对位置信息的感知度和敏感度。为了对样本进行更为精确的平衡处理，引入Focal Loss损失函数，进一步提高模型的检测精度与鲁棒性。实验结果表明，在景区行人数据集上，改进后的模型相较于原始模型，模型参数量减小了52%，mAP@0.5提升了2.1个百分点，mAP@0.5：0.95提升了1.4个百分点。在VisDrone2019数据集上，mAP@0.5提高了3.9个百分点。改进后的算法具有更强的泛化性能，能够更好地适用于景区行人检测任务。

关键词: 行人检测, 轻量化, YOLOv8, Focal Loss, 注意力机制

Abstract: Aiming at the problems of large pedestrian flow and dense crowds in scenic spots and the low efficiency of existing target detection algorithms for detecting occluded targets and small targets and the large number of model parameters, a lightweight scenic pedestrian detection algorithm SSC-YOLOv8n based on YOLOv8n is proposed. Firstly, the spatial and channel reconstruction attention convolution SCC2fEMA module is proposed to significantly reduce the number of model parameters and thereby improve the detection speed of the model. Secondly, the refined slim-neck paradigm is adopted, and the GSConv and V0V-GSCSP modules are used to effectively reduce the number of model parameters while improving the learning ability of the model. In addition, a coordinate attention dynamic decoupling head is proposed to significantly enhance the model’s perception and sensitivity to position information. Finally, in order to more accurately balance the samples, the Focal Loss function is introduced to further improve the detection accuracy and robustness of the model. Experimental results show that on the scenic pedestrian data set, compared with the original model, the improved model is reduced the number of model parameters by 52%, mAP@0.5 is increased by 2.1 percentage poins, and mAP@0.5：0.95 is increased by 1.4 percentage poins. It shows that on the VisDrone2019 data set, mAP@0.5 is increased by 3.9percentage points. The improved algorithm has stronger generalization performance and can be better suitable for pedestrian detection tasks in scenic spots.

Key words: pedestrian detection, lightweight, YOLOv8, Focal Loss, attention mechanism

张小艳, 王苗. 改进的YOLOv8n轻量化景区行人检测方法研究[J]. 计算机工程与应用, 2025, 61(2): 84-96.

ZHANG Xiaoyan, WANG Miao. Research on Improved YOLOv8n Light-Weight Pedestrian Detection Method in Scenic Spots[J]. Computer Engineering and Applications, 2025, 61(2): 84-96.

参考文献

[1] ARINDAM C. Smart traffic management of vehicles using faster R-CNN based deep learning method[J]. Scientific Reports, 2024, 14: 10357.
[2] BAI T, LUO J, ZHOU S, et al. Vehicle-type recognition method for images based on improved faster R-CNN model[J]. Sensors, 2024, 24(8): 2650.
[3] 贺艺斌, 田圣哲, 兰贵龙. 基于改进Faster-RCNN算法的行人检测[J]. 汽车实用技术, 2022, 47(5): 34-37.
HE Y B, TIAN S Z, LAN G L. Pedestrian detection based on improved Faster-RCNN algorithm[J]. Automobile Applied Technology, 2022, 47(5): 34-37.
[4] DIWAN T, ANIRUDH G, TEMBHURNE J V. Object detection using YOLO: challenges, architectural successors, datasets and applications[J]. Multimedia Tools and Applications, 2023, 82(6): 9243-9275.
[5] YANG Y, CHEN P, DING K, et al. Object detection of inland waterway ships based on improved SSD model[J]. Ships and Offshore Structures, 2023, 18(8): 1192-1200.
[6] 王莹, 田莹. 基于改进YOLOv5s的复杂环境行人检测模型[J]. 微电子学与计算机, 2024, 41(3): 29-36.
WANG Y, TIAN Y. Pedestrian detection model in complex environment based on improved YOLOv5s[J]. Microelectronics & Computer, 2024, 41(3): 29-36.
[7] 刘嘉泽, 王超, 生龙. 基于YOLOv5的行人检测方法研究[J]. 电脑与信息技术, 2024, 32(1): 37-41.
LIU J Z, WANG C, SHENG L. Research on pedestrian detection method based on YOLOv5[J]. Computer and Information Technology, 2024, 32(1): 37-41.
[8] TAHIR N U A, LONG Z, ZHANG Z, et al. PVswin-YOLOv8s: UAV-based pedestrian and vehicle detection for traffic management in smart cities using improved YOLOv8[J]. Drones, 2024, 8(3): 84.
[9] LI J, WEN Y, HE L. SCConv: spatial and channel reconstruction convolution for feature redundancy[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023: 6153-6162.
[10] OUYANG D, HE S, ZHANG G, et al. Efficient multi-scale attention module with cross-spatial learning[C]//Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing, 2023: 1-5.
[11] HOU Q, ZHOU D, FENG J. Coordinate attention for efficient mobile network design[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021: 13713-13722.
[12] LI H, LI J, WEI H, et al. Slim-neck by GSConv: a better design paradigm of detector architectures for autonomous vehicles[J]. arXiv: 2206.02424, 2022.
[13] LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection[C]//Proceedings of the IEEE International Conference on Computer Vision, 2017: 2980-2988.
[14] LI X, WANG W, HU X, et al. Selective kernel networks[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019: 510-519.
[15] ZHANG D, ZHENG Z, LI M, et al. CSART: channel and spatial attention-guided residual learning for real-time object tracking[J]. Neurocomputing, 2021, 436: 260-272.
[16] 盛帅, 段先华, 胡维康, 等. Dynamic-YOLOX: 复杂背景下的苹果叶片病害检测模型[J]. 计算机科学与探索, 2024, 18(8): 2118-2129.
SHENG S, DUAN X H, HU W K, et al. Dynamic-YOLOX: detection model for apple leaf disease in complex background[J]. Journal of Frontiers of Computer Science and Technology, 2024, 18(8): 2118-2129.
[17] 储珺, 束雯, 周子博, 等. 结合语义和多层特征融合的行人检测[J]. 自动化学报, 2022, 48(1): 282-291.
CHU J, SHU W, ZHOU Z B, et al. Combining semantics with multi-level feature fusion for pedestrian detection[J]. Acta Automatica Sinica, 2022, 48(1): 282-291.
[18] GE Z, LIU S, WANG F, et al. Yolox: exceeding yolo series in 2021[J]. arXiv: 2107. 08430, 2021.
[19] 颜豪男, 吕伏, 冯永安. 特征级自适应增强的无人机目标检测算法[J]. 计算机科学与探索, 2024, 18(6): 1566-1578.
YAN H N, LYU F, FENG Y A. Feature-level adaptive enhancement for UAV target detection algorithm[J]. Journal of Frontiers of Computer Science and Technology, 2024, 18(6): 1566-1578.
[20] CAI T, ZHANG D, WANG Y, et al. Learning local-global feature representation for pedestrian detection and re-identification[C]//Proceedings of the 2023 7th Asian Conference on Artificial Intelligence Technology, 2023: 756-763.
[21] CHOLLET F. Xception: deep learning with depthwise separable convolutions[C]//Proceedings of the IEEE conference on computer vision and pattern recognition, 2017: 1251-1258.
[22] REIS D, KUPEC J, HONG J, et al. Real-time flying object detection with YOLOv8[J]. arXiv: 2305.09972, 2023.
[23] HOWARD A, SANDLER M, CHU G, et al. Searching for mobilenetv3[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019: 1314-1324.
[24] LIU X, PENG H, ZHENG N, et al. EfficientViT: memory efficient vision transformer with cascaded group attention[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023: 14420-14430.
[25] ZHANG X, ZHOU X, LIN M, et al. ShuffleNet: an extremely efficient convolutional neural network for mobile devices[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 6848-6856.
[26] HAN K, WANG Y, TIAN Q, et al. GhostNet: more features from cheap operations[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020: 1580-1589.