基于改进YOLOX-s的车辆检测方法研究

doi:10.16182/j.issn1004731x.joss.22-1106

系统仿真学报 ›› 2024, Vol. 36 ›› Issue (2): 487-496.doi: 10.16182/j.issn1004731x.joss.22-1106

基于改进YOLOX-s的车辆检测方法研究

张稀柳(), 张晓玲(), 何敏军

江苏理工学院电气信息工程学院，常州 213001

收稿日期:2022-09-20 修回日期:2022-12-08 出版日期:2024-02-15 发布日期:2024-02-04
通讯作者: 张晓玲 E-mail:zxl13606146329@163.com;23175809@qq.com
第一作者简介:张稀柳(1998-)，女，硕士生，研究方向为机器视觉、图像处理。E-mail：zxl13606146329@163.com
基金资助:
2021年江苏理工学院研究生实践创新项目(XSJCX21_57)

Research on Vehicle Detection Method Based on Improved YOLOX-s

Zhang Xiliu(), Zhang Xiaoling(), He Minjun

College of Electrical and Information Engineering, Jiangsu University of Technology, Changzhou 213001, China

Received:2022-09-20 Revised:2022-12-08 Online:2024-02-15 Published:2024-02-04
Contact: Zhang Xiaoling E-mail:zxl13606146329@163.com;23175809@qq.com

摘要/Abstract

摘要：

为缓解车辆小目标漏检及误检问题，提出一种基于YOLOX网络的多尺度特征融合的改进车辆检测模型。设计基于深度可分离卷积的Ghost-CSP（cross stage partial），替换网络的部分跨阶段局部结构，加快检测速度；将模型的最大池化方式改进为Softpool方式，并引入坐标注意力机制，增强待检测目标的特征表达，优化目标漏检问题；选用Focal Loss作为模型置信度损失函数以增加分类不准确样本的权重，提高模型对小目标的预测能力。实验结果表明：改进算法平均准确率提高到74.96%，速度达到73帧/s，在满足实时性要求下可以更好地完成车辆目标检测要求。

关键词: YOLOX, 多尺度特征融合, 车辆检测模型, Softpool, 坐标注意力, Focal Loss

Abstract:

A improved vehicle detection model based on multi-scale feature fusion of YOLOX network is proposed to solve the problem of missing and false detection of small vehicle targets. Ghost-cross stage partial(CSP) based on the depth separable convolution is designed to replace part of cross stage partial in network to speed up the speed of detection. The max pooling mode of model is improved to Softpool mode, and coordinate attention mechanism is introduced to enhance the feature expression of target to be detected and to optimize the problem of target missing detection. Focal Loss is selected as the confidence loss function of model to increase the weight of inaccurate classification samples and improve the prediction ability of the model for small targets. The experimental results show that the average accuracy of the improved algorithm is improved to 74.96%, and the speed is up to 73 frames per second, which can better meet the requirements of real-time vehicle target detection.

Key words: YOLOX, multi-scale feature fusion, vehicle detection model, Softpool, coordinate attention, Focal Loss

中图分类号:

TP391

张稀柳,张晓玲,何敏军 . 基于改进YOLOX-s的车辆检测方法研究[J]. 系统仿真学报, 2024, 36(2): 487-496.

Zhang Xiliu,Zhang Xiaoling,He Minjun . Research on Vehicle Detection Method Based on Improved YOLOX-s[J]. Journal of System Simulation, 2024, 36(2): 487-496.

图/表 11

图1

图2

图3

图4

图5

图6

图7

图8

表1

表2

图9

参考文献 24

1	桑振. 基于单目视觉的前方车辆测距测速方法研究[D]. 北京: 北京交通大学, 2020.
	Sang Zhen. Study on Distance and Velocity Measurement Method of Front Vehicle Based on Monocular Vision[D]. Beijing: Beijing Jiaotong University, 2020.
2	Chen Duanyu, Lin Yuhao, Peng Yangjie. Nighttime Brake-light Detection by Nakagami Imaging[J]. IEEE Transactions on Intelligent Transportation Systems, 2012, 13(4): 1627-1637.
3	朱英凯, 罗文广, 宾洋. 基于改进车底阴影提取算法的前方运动车辆实时检测[J]. 电子技术应用, 2018, 44(4): 86-89, 98.
	Zhu Yingkai, Luo Wenguang, Yang Bin. Real-time Detection of Front Moving Vehicles Based on Improved Vehicle Shadow Extraction Algorithm[J]. Application of Electronic Technique, 2018, 44(4): 86-89, 98.
4	陈泽, 叶学义, 钱丁炜, 等. 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.
	Chen Ze, Ye Xueyi, Qian Dingwei, et al. Small-scale Pedestrian Detection Based on Improved Faster R-CNN[J]. Computer Engineering, 2020, 46(9): 226-232, 241.
5	茅智慧, 朱佳利, 吴鑫, 等. 基于YOLO的自动驾驶目标检测研究综述[J]. 计算机工程与应用, 2022, 58(15): 68-77.
	Mao Zhihui, Zhu Jiali, Wu Xin, et al. Review of YOLO Based Target Detection for Autonomous Driving[J]. Computer Engineering and Applications, 2022, 58(15): 68-77.
6	Gao Xue, Han Siyi, Luo Cong. A Detection and Verification Model Based on SSD and Encoder-decoder Network for Scene Text Detection[J]. IEEE Access, 2019, 7: 71299-71310.
7	Redmon J, Divvala S, Girshick R, et al. You Only Look Once: Unified, Real-time Object Detection[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, NJ, USA: IEEE, 2016: 779-788.
8	Redmon J, Farhadi A. YOLO9000: Better, Faster, Stronger[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, NJ, USA: IEEE, 2017: 6517-6525.
9	Mao Qichao, Sun Hongmei, Liu Yanbo, et al. Mini-YOLOv3: Real-time Object Detector for Embedded Applications[J]. IEEE Access, 2019, 7: 133529-133538.
10	於积荣, 黄德启, 曾蓉, 等. 基于改进YOLOv4的车型检测算法研究[J]. 激光杂志, 2022, 43(4): 52-59.
	Yu Jirong, Huang Deqi, Zeng Rong, et al. Research on Vehicle Detection Algorithm Based on Improved YOLOv4[J]. Laser Journal, 2022, 43(4): 52-59.
11	金立生, 郭柏苍, 王芳荣, 等. 基于改进YOLOv3的车辆前方动态多目标检测算法[J]. 吉林大学学报(工学版), 2021, 51(4): 1427-1436.
	Jin Lisheng, Guo Baicang, Wang Fangrong, et al. Dynamic Multiple Object Detection Algorithm for Vehicle Forward Based on Improved YOLOv3[J]. Journal of Jilin University(Engineering and Technology Edition), 2021, 51(4): 1427-1436.
12	柳长源, 王琪, 毕晓君. 多目标小尺度车辆目标检测方法[J]. 控制与决策, 2021, 36(11): 2707-2712.
	Liu Changyuan, Wang Qi, Bi Xiaojun. Multi-target and Small-scale Vehicle Target Detection Method[J]. Control and Decision, 2021, 36(11): 2707-2712.
13	赵家琪, 高贵, 黄晓峰, 等. 一种改进的YOLOv4-tiny车辆目标检测方法[J]. 电子产品世界, 2021, 28(9): 39-43.
	Zhao Jiaqi, Gao Gui, Huang Xiaofeng, et al. An Improved YOLOv4-tiny Vehicle Target Detection Method[J]. Electronic Engineering & Product World, 2021, 28(9): 39-43.
14	Ge Zheng, Liu Songtao, Wang Feng, et al. YOLOX: Exceeding YOLO Series in 2021[EB/OL]. (2021-08-06) [2022-03-17]. .
15	Wang Ling, Zhou Hai, Bian Chunjiang, et al. Hardware Acceleration and Implementation of YOLOX-s for On-Orbit FPGA[J]. Electronics, 2022, 11(21): 3473.
16	熊怡梦. 基于标签语义和Transformer的元学习小样本目标检测方法研究[D]. 西安: 西安电子科技大学, 2021.
	Xiong Yimeng. Label Semantics and Transformer for Meta Learning Few-shot Object Detection[D]. Xi'an: Xidian University, 2021.
17	Mercedes E Paoletti, Haut Juan M, Pereira Nuno S, et al. Ghostnet for Hyperspectral Image Classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 59(12): 10378-10393.
18	史芳菲. 监控视频下运营车辆司机吸烟行为检测系统的设计与实现[D]. 海口: 海南大学, 2021.
	Shi Fangfei. Design and Implementation of a Smoking Detection System for Operating Vehicle Drivers Under Surveillance Video[D]. Haikou: Hainan University, 2021.
19	He Kaiming, Zhang Xiangyu, Ren Shaoqing, 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.
20	Wang Chunyuan, Wu Yang, Wang Yihan, et al. Scene Recognition Using Deep Softpool Capsule Network Based on Residual Diverse Branch Block[J]. Sensors, 2021, 21(16): 5575.
21	曹桢淼, 吉卫喜, 苏璇, 等. 基于扩展跨阶段局部网络的表面缺陷检测[J]. 计算机集成制造系统, 2022, 28(8): 2399-2407.
	Cao Zhenmiao, Ji Weixi, Su Xuan, et al. Surface Defect Detection Based on Scaling Cross-stage Partial Network[J]. Computer Integrated Manufacturing Systems, 2022, 28(8): 2399-2407.
22	Xie Chao, Zhu Hongyu, Fei Yeqi. Deep Coordinate Attention Network for Single Image Super-resolution[J]. IET Image Processing, 2021, 16(1): 273-284.
23	Liu Zechuan, Wang Song. Broken Corn Detection Based on an Adjusted YOLO with Focal Loss[J]. IEEE Access, 2019, 7: 68281-68289.
24	史彩娟, 涂冬景, 刘靖祎. Re-GAN: 残差生成式对抗网络算法[J]. 中国图象图形学报, 2021, 26(3): 594-604.
	Shi Caijuan, Tu Dongjing, Liu Jingyi. Re-GAN: Residual Generative Adversarial Network Algorithm[J]. Journal of Image and Graphics, 2021, 26(3): 594-604.

Multi-scale fusion	Ghost-CSP	SoftPool	CA	Focal Loss	mAP50/%	FPS/(帧/s)	模型大小/MB
—	—	—	—	—	70.94	77	34.6
√	—	—	—	—	72.63	72	35.5
—	√	—	—	—	71.08	79	30.8
—	—	√	√	—	72.96	73	36.9
—	—	—	—	√	71.11	77	35.0
√	√	√	√	√	74.96	73	34.1

Network model	AP50/%			召回/%	mAP/%	FPS/(帧/s)
Network model	Car	Bus	Truck	召回/%	mAP/%	FPS/(帧/s)
YOLOv3	68.32	57.83	58.46	55.26	61.54	27
YOLOv4	72.23	60.78	61.94	59.08	64.98	48
YOLOv5s	77.31	62.67	63.08	61.73	67.69	69
YOLOX-s	83.12	64.43	65.26	63.44	70.94	77
Improved YOLOX-s	88.50	67.71	68.67	66.18	74.96	73

基于改进YOLOX-s的车辆检测方法研究

Research on Vehicle Detection Method Based on Improved YOLOX-s

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 11

参考文献 24

相关文章 1

编辑推荐

Metrics

本文评价