文章编号: 68125
文献标识码: A
基于改进YOLOv8的景区行人检测算法
, {{javascript:window.splitNeed=true;}}
, {{custom_author.xref}}
,
收稿日期：2023-07-21
网络出版日期：2023-12-19
基金资助
甘肃省重点研发计划-工业类项目(22YF7GA159)
甘肃省基础研究计划-软科学专项(22JR4ZA084)
甘肃省教育厅产业支撑计划项目(2023CYZC-25)
{{custom_fund}}
版权
Pedestrian Detection Algorithm for Scenic Spots Based on Improved YOLOv8
, {{javascript:window.splitNeed=true;}}
, {{custom_author.xref_en}}
,
Received：2023-07-21
Online：2023-12-19
Copyright
针对当前景区行人检测具有检测精度低、算法参数量大和现有公开数据集在小目标检测上存在限制等问题, 创建TAPDataset行人检测数据集, 弥补现有数据集在小目标检测方面的不足, 并基于YOLOv8算法, 构建一种检测精度高、硬件要求低的新模型YOLOv8-L。首先引入DepthSepConv轻量化卷积模块, 降低模型的参数量和计算量。其次采用BiFormer注意力机制和上采样算子CARAFE, 加强模型对图像的语义理解和信息融合能力, 提升模型的检测精度。最后增加一层小目标检测层来提取更多的浅层特征, 从而有效地改善模型对小目标的检测性能。在TAPDataset、VOC 2007及TAP+VOC数据集上的实验结果表明, 与YOLOv8相比, 在FPS基本不变的情况下, 在TAPDataset数据集上, 模型的参数量减少了18.06%, mAP@0.5提高了5.51%, mAP@0.5∶0.95提高了6.03%;在VOC 2007数据集上, 模型的参数量减少了13.6%, mAP@0.5提高了3.96%, mAP@0.5∶0.95提高了6.39%;在TAP+VOC数据集上, 模型的参数量减少了14.02%, mAP@0.5提高了4.49%, mAP@0.5∶0.95提高了5.68%。改进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。
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. DOI: 10.19678/j.issn.1000-3428.0068125
Xiangquan GUI, Shiqing LIU, Li LI, Qingsong QIN, Tangyan LI. Pedestrian Detection Algorithm for Scenic Spots Based on Improved YOLOv8[J]. Computer Engineering, 2024, 50(7): 342-351. DOI: 10.19678/j.issn.1000-3428.0068125
The TAPDataset pedestrian detection dataset is used in this study to address the issues of low detection accuracy, large number of algorithm parameters, and limitations of existing public datasets for small target detection in current scenic pedestrian detection. This dataset addresses the deficiencies of existing datasets regarding small target detection. Based on the YOLOv8 algorithm, a new model with high detection accuracy and low hardware requirements, called YOLOv8-L, is proposed. First, the lightweight convolution module DepthSepConv is introduced to reduce the number of parameters and computations of the model. Second, the BiFormer attention mechanism and CARAFE upsampling operator are used to enhance the model's semantic understanding of images and information fusion capability, significantly improving detection accuracy. Finally, a small target detection layer is added to extract more shallow features, effectively improving the model's performance for small target detection. The effectiveness of the algorithm is verified using the TAPDataset, VOC 2007, and TAP+VOC datasets. The experimental results show that compared with YOLOv8, the number of model parameters is reduced by 18.06% on the TAPDataset with unchanged FPS, mAP@0.5 improves by 5.51%, and mAP@0.5∶0.95 improves by 6.03%. On the VOC 2007 dataset, the number of parameters is reduced by 13.6%, with mAP@0.5 improving by 3.96% and mAP@0.5∶0.95 improving by 6.39%. On the TAP+VOC dataset, the number of parameters is reduced by 14.02%, with mAP@0.5 improving by 4.49% and mAP@0.5∶0.95 improving by 5.68%. The improved algorithm demonstrates stronger generalization performance and can be better applied to scenic pedestrian detection tasks.
表2 各种注意力机制在YOLOv8下的对比Table 2 Comparison of various attention mechanisms under YOLOv8 |
 |
表3 增加小目标检测层后与原模型的实验对比Table 3 Experimental comparison with the original model after adding small target detection layer |
 |
表4 数据集划分结果Table 4 Results of dataset segmentation |
 |
表5 实验环境Table 5 Experimental environment |
 |
表6 YOLOv8-L在TAPDataset上的对比实验Table 6 Comparative experiments of YOLOv8-L on TAPDataset |
 |
表7 YOLOv8与YOLOv8-L在不同数据集上的对比实验Table 7 Comparative experiments of YOLOv8 and YOLOV8-L on different datasets |
 |
表8 YOLOv8-L在VOC 2007数据集上的对比实验Table 8 Comparative experiments of YOLOv8-L on VOC 2007 dataset |
 |
表9 消融实验结果Table 9 Ablation experiment results |
 |
| 1 |
TANG C, HOU C P. RGBD salient object detection by structured low-rank matrix recovery and Laplacian constraint. Transactions of Tianjin University, 2017, 23(2): 176- 183.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 2 |
王亮, 张超. 一种基于YOLOv5的轻量型行人检测方法. 工业控制计算机, 2023, 36(4): 84-86, 89.
WANG L, ZHANG C. A lightweight pedestrian detection method based on YOLOv5. Industrial Control Computer, 2023, 36(4): 84-86, 89.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 3 |
ZOU Z X, CHEN K Y, SHI Z W, et al. Object detection in 20 years: a survey[EB/OL]. [2023-06-10]. https://levir.buaa.edu.cn/publications/od_survey.pdf.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 4 |
洪松, 高定国. 基于YOLOv3的车辆和行人检测方法. 电脑知识与技术, 2020, 16(8): 192-193, 198.
HONG S, GAO D G. Vehicle and pedestrian detection method based on YOLOv3. Computer Knowledge and Technology, 2020, 16(8): 192-193, 198.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 5 |
HONG W W, MA Z H, et al. Detection of green asparagus in complex environments based on the improved YOLOv5 algorithm. Sensors, 2023, 23, 1562.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 6 |
WANG Q, FENG W Q, YAO L F, et al. TPH-YOLOv5-air: airport confusing object detection via adaptively spatial feature fusion. Remote Sensing, 2023, 15(15): 3883.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 7 |
TERVEN J, CORDOVA-ESPARZA D. A comprehensive review of YOLO: from YOLOv1 to YOLOv8 and beyond[EB/OL]. [2023-06-10]. https://arxiv.org/pdf/2304.00501.pdf.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 8 |
邱天衡, 王玲, 王鹏, 等. 基于改进YOLOv5的目标检测算法研究. 计算机工程与应用, 2022, 58(13): 63- 73.
QIU T H, WANG L, WANG P, et al. Research on object detection algorithm based on improved YOLOv5. Computer Engineering and Applications, 2022, 58(13): 63- 73.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 9 |
孙传猛, 王燕平, 王冲, 等. 融合改进YOLOv3与三次样条插值的煤岩界面识别方法. 采矿与岩层控制工程学报, 2022, 4(1): 81- 90.
SUN C M, WANG Y P, WANG C, et al. Coal-rock interface identification method based on improved YOLOv3 and cubic spline interpolation. Journal of Mining and Strata Control Engineering, 2022, 4(1): 81- 90.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 10 |
王程, 刘元盛, 刘圣杰. 基于改进YOLOv4的小目标行人检测算法. 计算机工程, 2023, 49(2): 296-302, 313.
WANG C, LIU Y S, LIU S J. Small-target pedestrian-detection algorithm based on improved YOLOv4. Computer Engineering, 2023, 49(2): 296-302, 313.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 11 |
严开忠, 马国梁, 许立松, 等. 基于改进YOLOv3的机载平台目标检测算法. 电光与控制, 2021, 28(5): 70- 74.
YAN K Z, MA G L, XU L S, et al. Improved YOLOv3 based target detection algorithm for airborne platform. Electronics Optics & Control, 2021, 28(5): 70- 74.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 12 |
王艺成, 张国良, 张自杰. 基于改进YOLOv5的小目标检测方法. 计算机与现代化, 2023,(5): 100- 105.
WANG Y C, ZHANG G L, ZHANG Z J. Small object detection method based on improved YOLOv5. Computer and Modernization, 2023,(5): 100- 105.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 13 |
李文豪, 周斌, 胡波, 等. 基于轻量化网络的遮挡人脸检测. 中南民族大学学报(自然科学版), 2022, 41(3): 339- 346.
LI W H, ZHOU B, HU B, et al. Occlusion face detection based on lightweight network. Journal of South-Central Minzu University (Natural Science Edition), 2022, 41(3): 339- 346.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 14 |
李闻, 李小春, 闫昊雷. 基于改进YOLO v3的PCB缺陷检测. 电光与控制, 2022, 29(4): 106- 111.
LI W, LI X C, YAN H L. PCB defect detection based on improved YOLO v3. Electronics Optics & Control, 2022, 29(4): 106- 111.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 15 |
CUI C, GAO T Q, WEI S Y, et al. PP-LCNet: a lightweight CPU convolutional neural network[EB/OL]. [2023-06-10]. http://arxiv.org/abs/2109.15099v1.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 16 |
ZHAN Y, YU J. Multi-task compositional network for visual relationship detection. International Journal of Computer Vision, 2020, 128(8): 2146- 2165.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 17 |
ZHU L, WANG X J, KE Z G, et al. BiFormer: vision transformer with bi-level routing attention[EB/OL]. [2023-06-10]. https://arxiv.org/abs/2303.08810.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 18 |
杨永波, 李栋. 改进YOLOv5的轻量级安全帽佩戴检测算法. 计算机工程与应用, 2022, 58(9): 201- 207.
YANG Y B, LI D. Lightweight helmet wearing detection algorithm of improved YOLOv5. Computer Engineering and Applications, 2022, 58(9): 201- 207.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 19 |
WANG J Q, CHEN K, XU R, et al. CARAFE: content-aware ReAssembly of FEatures[C]//Proceedings of IEEE/CVF International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2019: 458-467.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 20 |
WU D L, JIANG S, ZHAO E L, et al. Detection of camellia oleifera fruit in complex scenes by using YOLOv7 and data augmentation. Applied Sciences, 2022, 12(22): 11318.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 21 |
ABOAH A, WANG B, BAGCI U, et al. Real-time multi-class helmet violation detection using few-shot data sampling technique and YOLOv8[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2023: 5350-5358.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 22 |
GEOFFERY A. Real-time helmet violation detection using YOLOv5 and ensemble learning[EB/OL]. [2023-06-10]. https://arxiv.org/pdf/2304.09246.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 23 |
LAWAL O M. YOLOv5-LiNet: a lightweight network for fruits instance segmentation. PLoS One, 2023, 18(3): e0282297.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 24 |
KUMAR S, KUMAR A, KUMAR K. GhostNet-YOLO algorithm for object detection in UAV image[C]//Proceedings of the 7th International Conference on Image Information Processing. Washington D. C., USA: IEEE Press, 2023: 293-299.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| 25 |
WANG C Y, BOCHKOVSKIY A, LIAO H Y M. YOLOv7: trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2023: 2781-2795.
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
| {{custom_reference.label}} |
{{custom_citation.content}}
{{javascript:window.needSplit=false;}}
{{javascript:window.needSplit=true;}}DOI {{javascript:window.needSplit=true;}}URL PMID
{{custom_citation.annotation}}
|
图1 YOLOv8架构图2 Conv模块结构图3 C2f模块结构图4 SPPF结构
表1 YOLOv8-L网络结构图5 YOLOv8-L架构图6 深度卷积图7 逐点卷积图8 BRA结构图9 BiFormer结构表2 各种注意力机制在YOLOv8下的对比表3 增加小目标检测层后与原模型的实验对比表4 数据集划分结果表5 实验环境表6 YOLOv8-L在TAPDataset上的对比实验表7 YOLOv8与YOLOv8-L在不同数据集上的对比实验表8 YOLOv8-L在VOC 2007数据集上的对比实验图10 YOLOv5、YOLOv8和YOLOv8-L效果对比表9 消融实验结果/
| 〈 |
|
〉 |
