基于YOLOv4改进算法的复杂行人检测模型研究

计算机工程与科学 ›› 2022, Vol. 44 ›› Issue (08): 1449-1456.

基于YOLOv4改进算法的复杂行人检测模型研究

李兰,刘杰,张洁

(青岛理工大学信息与控制工程学院，山东青岛 266000)

收稿日期:2020-12-30 修回日期:2021-02-25 接受日期:2022-08-25 出版日期:2022-08-25 发布日期:2022-08-25

A complex pedestrian detection model based on improved YOLOv4 algorithm

LI Lan,LIU Jie,ZHANG Jie

(School of Information and Control Engineering,Qingdao University of Technology,Qingdao 266000,China)

Received:2020-12-30 Revised:2021-02-25 Accepted:2022-08-25 Online:2022-08-25 Published:2022-08-25

摘要/Abstract

摘要： 在图像和视频序列中进行行人检测时，存在行人姿态和尺度多样及行人遮挡的问题，导致YOLOv4算法对部分行人检测不准确，存在误检和漏检的情况。针对这一问题，提出了基于YOLOv4改进算法的复杂行人检测模型。首先，使用改进的k-means聚类算法对行人数据集真实框尺寸进行分析，根据聚类结果确定先验框尺寸；其次，利用PANet进行多尺度特征融合，增强对多姿态、多尺度行人目标的敏感度，以提高检测效果；最后，针对行人遮挡问题，使用斥力损失函数使预测框尽可能地靠近正确的目标。实验表明，相比于YOLOv4和其他行人检测模型，新提出的检测模型具有更好的检测效果。

关键词: 行人检测, YOLOv4算法, k-means聚类算法, 多尺度特征融合, 斥力损失

Abstract: When pedestrian detection is carried out in image and video sequences, there are some problems, such as diverse pedestrian posture and scale as well as pedestrian occlusion, which leads to inaccurate detection of some pedestrians by YOLOv4 algorithm and false detection and missed detection. In order to solve this problem, a complex pedestrian detection model based on improved YOLOv4 algorithm is proposed. Firstly, the ground truth size of pedestrian dataset is analyzed by the improved k-means clustering algorithm, and the size of anchor box is determined according to the clustering results. Secondly, PANet is used for multi-scale feature fusion to make it more sensitive to multi-attitude and multi-scale pedestrian targets and improve the detection effect. Finally, for pedestrian occlusion problem, the repulsion loss function is proposed to make the predicted box as close to the correct target as possible. The experimental results show that the new de-tection model has better detection effect than YOLOv4 and other pedestrian detection model.

Key words: pedestrian detection, YOLOv4 algorithm, k-means clustering algorithm, multi-scale feature fusion, repulsion loss

李兰, 刘杰, 张洁. 基于YOLOv4改进算法的复杂行人检测模型研究[J]. 计算机工程与科学, 2022, 44(08): 1449-1456.

LI Lan, LIU Jie, ZHANG Jie. A complex pedestrian detection model based on improved YOLOv4 algorithm[J]. Computer Engineering & Science, 2022, 44(08): 1449-1456.

参考文献［8］

［1］	Song Yan-yan,Tan Li,Ma Zi-hao,et al.Video target detection based on improved YOLOv3 algorithm［J］.Journal of Frontiers of Computer Science and Technology,2021,15(1):163-172.(in Chinese)
［2］	Wang Dian-wei, Yang Xu,Han Peng-fei,et al.Panoramic vi- deo motion small target detection algorithm in complex background［J］.Control and Decision,2021,36(1):249-256.(in Chinese)
［3］	Kong Fang-fang,Song Bei-bei.Improved YOLOv3 panoramic traffic monitoring target detection［J］.Computer Engineering and Applications,2020,56(8):20-25.(in Chinese)
［4］	Bodla N,Singh B,Chellappa R,et al.Improving object detection with one line of code［C］∥Proc of the IEEE International Conference on Computer Vision,2017:5561-5569.
［5］	Zheng Z H,Wang P,Liu W,et al.Distance-IoU loss:Faster and better learning for bounding box regression［C］∥Proc of the AAAI Conference on Artificial Intelligence,2020:12993-13000.
［6］	Girshick R,Donahue J,Darrell T,et al.Rich feature hierarchies for object detection and semantic segmentation［C］∥Proc of the IEEE Conference on Computer Vision and Pattern Recognition,2014:580-587.
［7］	Cai Jia-ran.Research on pedestrian detection based on deep learning in complex background［D］.Guangzhou:South China University of Technology,2019.(in Chinese)
［8］	Luo Peng-fei. Research on pedestrian detection and action recognition based on deep learning［D］.Hefei:University of Science and Technology of China,2020.(in Chinese)
［9］	Girshick R.Fast R-CNN［C］∥Proc of the IEEE International Conference on Computer Vision,2015:1440-1448.
［10］	Ren S, He K, Girshick R,et al.Faster R-CNN:Towards real-time object detection with region proposal networks［C］∥Proc of Internatioanl Conference on Neural Information Processing Systems,2015:91-99.
［11］	Dai J F, Li Y ,He K M,et al.R-FCN:Object detection via region-based fully convolutional networks［C］∥Proc of Internatioanl Conference on Neural Information Processing Systems,2016:379-387.
［12］	Redmon J,Divvala S,Girshick R,et al.You only look once:Unified,real-time object detection［C］∥Proc of the IEEE Conference on Computer Vision and Pattern Recognition,2016:779-788.
［13］	Redmon J,Farhadi A.YOLO9000:Better,faster,stronger［C］∥Proc of the IEEE Conference on Computer Vision and Pattern Recognition,2017:7263-7271.
［14］	Redmon J,Farhadi A.YOLOv3:An incremental improvement［J］.arXiv:1804.02767,2018.
［15］	Bochkovskiy A,Wang C,Liao H.YOLOv4:Optimal speed and accuracy of object detection［J］.arXiv:2004.10934,2020.
［16］	He K M,Zhang X Y,Ren S Q,et al.Spatial pyramid pooling in deep convolutional networks for visual recognition［J］.IEEE Transactions on Pattern Analysis and Machine Intelligence,2015,37(9):1904-1916.
［17］	Liu S,Qi L,Qin H F,et al.Path aggregation network for instance segmentation［C］∥Proc of the IEEE/CVF Confe- rence on Computer Vision and Pattern Recognition,2018:8759-8768.
［18］	Wang X L,Xiao T T,Jiang Y N,et al.Repulsion loss:Detect- ing pedestrians in a crowd［C］∥Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,2018:7774-7783.
［19］	Zhang S S, Benenson R,Schiele B.CityPersons:A diverse dataset for pedestrian detection［C］∥Proc of the IEEE Conference on Computer Vision and Pattern Recognition,2017:249-256.
	附中文参考文献：
［1］	宋艳艳,谭励,马子豪,等.改进YOLOv3算法的视频目标检测［J］.计算机科学与探索,2021,15(1):163-172.
［2］	王殿伟,杨旭,韩鹏飞,等.复杂背景下全景视频运动小目标检测算法［J］.控制与决策,2021,36(1):249-256.
［3］	孔方方,宋蓓蓓.改进YOLOv3的全景交通监控目标检测［J］.计算机工程与应用,2020,56(8):20-25.
［7］	蔡佳然.复杂背景下基于深度学习的行人检测算法研究［D］.广州：华南理工大学,2019.
［8］	罗鹏飞.基于深度学习的行人检测与行为识别研究［D］.合肥：中国科学技术大学,2020.

[1]	曹玉东, 陈冬昊, 曹睿, 赵朗. 融合Mask R-CNN的在线多目标行人跟踪方法[J]. 计算机工程与科学, 2023, 45(07): 1216-1225.
[2]	崔克彬, 崔叶微. 基于卷积和Transformer的断路器动触头跟踪方法研究[J]. 计算机工程与科学, 2023, 45(07): 1236-1244.
[3]	孙琪, 翟锐, 左方, 张玉涛, . 基于部分卷积和多尺度特征融合的人脸图像修复模型[J]. 计算机工程与科学, 2023, 45(02): 304-312.
[4]	李利荣, 王子炎, 张开, 杨荻椿, 熊炜, 巩朋成, . 基于OSE-dResnet网络的列车底部零件检测算法[J]. 计算机工程与科学, 2022, 44(04): 692-698.
[5]	吴从中1,侯国松1,丁正龙2,许良凤1,詹曙1. 像素级的皮肤分割与面色分级[J]. 计算机工程与科学, 2019, 41(11): 1985-1990.
[6]	谢修娟1,李香菊1，莫凌飞2. 基于改进K-means算法的微博舆情分析研究[J]. 计算机工程与科学, 2018, 40(01): 155-158.