基于DYCS−YOLOv8n的井下无人驾驶电机车多目标检测

许谨辉; 王文善; 王爽; 王文钺; 赵婷婷

doi:10.13272/j.issn.1671-251x.2024100036

基于DYCS−YOLOv8n的井下无人驾驶电机车多目标检测

许谨辉^{1, 3,},
王文善^{1, 3, ,},
王爽^{1, 2, 3},
王文钺^{1, 3},
赵婷婷^{1, 3}

1.
安徽理工大学煤炭无人化开采数智技术全国重点实验室，安徽淮南　232001
2.
安徽理工大学矿山智能技术与装备省部共建协同创新中心，安徽淮南　232001
3.
安徽理工大学机电工程学院，安徽淮南　232001

基金项目:

国家自然科学基金面上项目（52274152）；安徽省智能矿山技术与装备工程研究中心开放基金项目（AIMTEERC202405）；安徽理工大学高层次引进人才科研启动基金项目（2023yjrc95）。

详细信息

作者简介:
许谨辉（2001—），男，安徽芜湖人，硕士研究生，研究方向为煤矿运输车视觉感知，E-mail：xujin1141@126.com

通讯作者:
王文善（1992—），男，安徽怀远人，讲师，博士，研究方向为煤矿无人驾驶运输车视觉感知，E-mail：wwsaust@126.com。

中图分类号: TD64
计量
- 文章访问数: 77
- HTML全文浏览量: 14
- PDF下载量: 10
出版历程
- 收稿日期: 2024-10-16
- 修回日期: 2025-04-22
- 网络出版日期: 2025-03-31
- 刊出日期: 2025-04-14

Multi-object detection for underground unmanned locomotives based on DYCS-YOLOv8n

XU Jinhui^{1, 3,},
WANG Wenshan^{1, 3, ,},
WANG Shuang^{1, 2, 3},
WANG Wenyue^{1, 3},
ZHAO Tingting^{1, 3}

1.
State Key Laboratory of Digital Intelligent Technology for Unmanned Coal Mining, Anhui University of Science and Technology, Huainan 232001, China
2.
Provincial and Ministerial Co-construction Collaborative Innovation Center for Intelligent Technology and Equipment of Mines, Anhui University of Science and Technology, Huainan 232001, China
3.
School of Mechatronics Engineering, Anhui University of Science and Technology, Huainan 232001, China

摘要

摘要:
针对井下无人驾驶电机车因光线暗、噪声大及运动模糊等因素导致图像特征难提取、细节易丢失、小尺寸目标难识别等问题，提出了一种基于DYCS−YOLOv8n的井下无人驾驶电机车多目标检测模型。在YOLOv8n的基础上引入卷积注意力模块（CBAM），通过空间和通道双重注意力机制，提高了对关键特征的提取能力；增加小目标检测层，由原来的3层增加到4层，从而更好地提取细小特征，提升了对小尺寸目标的检测性能；采用动态上采样算子DySample，根据输入特征自适应地调整采样策略，更好地保留图像中的边缘和局部细节，避免了图像关键信息损失。采用自建的井下无人驾驶电机车数据集进行实验，结果表明：① DYCS−YOLOv8n模型的平均精度均值(mAP@0.5)达97.5%，较YOLOv8n模型提高了3.4%，且检测速度达46.35帧/s，满足实时性检测需求。② 与YOLO系列主流目标检测模型相比，DYCS−YOLOv8n模型的mAP@0.5最优，在保持轻量化的同时保证了较快的计算速度。③ 在噪声、低光照等复杂井下场景下，DYCS−YOLOv8n模型对行人、轨道、信号灯的平均检测置信度较高，未出现漏与误检情况。
- 井下无人驾驶 /
- 电机车 /
- 多目标检测 /
- YOLOv8n /
- 卷积注意力机制 /
- 小目标检测 /
- 动态上采样
Abstract:
To address the challenges in underground unmanned locomotive image feature extraction—such as poor lighting, high noise, and motion blur, which result in the loss of image details and difficulty in identifying small targets—a multi-object detection model for underground unmanned locomotives based on DYCS-YOLOv8n was proposed. Based on YOLOv8n, the Convolutional Block Attention Module (CBAM) was introduced, enhancing the extraction of key features through spatial and channel attention mechanisms. A small-object detection layer was added, increasing the original three layers to four, thereby improving the extraction of fine features and enhancing detection performance for small-sized targets. The dynamic upsampling operator DySample was employed to adaptively adjust the sampling strategy according to the input features, better preserving edges and local details in the images and avoiding the loss of critical information. Experiments conducted on a self-constructed underground unmanned locomotive dataset showed that: ① The DYCS-YOLOv8n model achieved a mean Average Precision (mAP@0.5) of 97.5%, an improvement of 3.4% over the YOLOv8n model, with a detection speed of 46.35 frames per second, meeting the requirements for real-time detection. ② Compared with mainstream YOLO series object detection models, DYCS-YOLOv8n achieved the optimal mAP@0.5, maintaining a lightweight structure while ensuring high computational speed. ③ In complex underground scenarios with noise and low illumination, the DYCS-YOLOv8n model exhibited high average detection confidence for pedestrians, tracks, and signal lights, with no cases of missed or false detections.
- underground unmanned driving /
- locomotive /
- multi-object detection /
- YOLOv8n /
- convolutional attention mechanism /
- small-object detection /
- dynamic upsampling

HTML全文

图 1 DYCS−YOLOv8模型结构

Figure 1. Structure of DYCS-YOLOv8 model

下载: 全尺寸图片幻灯片

图 2 CBAM结构

Figure 2. CBAM structure

下载: 全尺寸图片幻灯片

图 3 通道注意力模块结构

Figure 3. Structure of channel attention module

下载: 全尺寸图片幻灯片

图 4 空间注意力模块结构

Figure 4. Structure of spatial attention module

下载: 全尺寸图片幻灯片

图 5 小目标检测层结构

Figure 5. Structure of small target detection layer

下载: 全尺寸图片幻灯片

图 6 DySample结构

Figure 6. DySample structure

下载: 全尺寸图片幻灯片

图 7 消融实验mAP@0.5曲线

Figure 7. mAP@0.5 curves of ablation experiment

下载: 全尺寸图片幻灯片

图 8 不同模型检测效果对比

Figure 8. Comparison of detection effect of different models

下载: 全尺寸图片幻灯片

表 1 不同注意力机制对比

Table 1 Comparison of different attention mechanisms

模型	AP/%			mAP@0.5/%	GFLOPs/10⁹
模型	person	track	light	mAP@0.5/%	GFLOPs/10⁹
YOLOv8n+SE	96.1	97.7	85.5	93.1	8.2
YOLOv8n+CA	96.0	97.1	87.1	93.4	8.2
YOLOv8n+EMA	96.3	97.2	89.4	94.3	8.3
YOLOv8n+GAM	96.4	97.0	85.3	92.9	9.5
YOLOv8n+CBAM	97.9	98.4	89.6	95.3	8.2

下载: 导出CSV

表 2 消融实验结果对比

Table 2 Comparison of ablation experiment results

模型	AP/%			mAP@0.5/%	GFLOPs/10⁹	Params/ 10⁶个	FPS/ （帧·s⁻¹）
模型	person	track	light	mAP@0.5/%	GFLOPs/10⁹	Params/ 10⁶个	FPS/ （帧·s⁻¹）
A	97.6	97.8	87.8	94.4	8.2	3.01	49.82
B	97.9	99.0	89.0	95.3	8.2	3.08	46.86
C	98.1	98.7	91.3	96.0	8.2	3.09	41.57
D	98.5	98.7	95.3	97.5	12.7	3.06	46.35

下载: 导出CSV

表 3 对比实验结果

Table 3 Comparative experiment results

模型	AP/%			mAP@0.5/%	Params/ 10⁶个	GFLOPs/ 10⁹
模型	person	track	light	mAP@0.5/%	Params/ 10⁶个	GFLOPs/ 10⁹
YOLOv5n	97.8	98.6	90.1	95.5	2.51	7.2
YOLOv6	98.0	98.1	87.7	94.6	4.20	11.8
YOLOv8s	95.6	98.4	87.7	93.9	11.13	28.6
YOLOv10n	97.9	98.5	89.8	95.4	2.71	8.2
YOLOv11n	98.2	98.7	89.6	95.5	2.60	6.4
DYCS−YOLOv8	98.5	98.7	95.3	97.5	3.06	12.7

下载: 导出CSV

参考文献(25)

[1]	崔邵云,鲍久圣,李芳威,等. 基于多源里程融合的井下无人驾驶自主导航SLAM方法[J/OL]. 煤炭科学技术:1-10[2024-10-06]. http://kns.cnki.net/kcms/detail/11.2402.td.20240925.1231.002.html. CUI Shaoyun,BAO Jiusheng,LI Fangwei,et al. Autonomous navigation SLAM method for underground unmanned driving based on multi-source mileage fusion scenarios[J/OL]. Coal Science and Technology:1-10[2024-10-06]. http://kns.cnki.net/kcms/detail/11.2402.td.20240925.1231.002.html.
[2]	杨豚,郭永存,王爽,等. 煤矿井下无人驾驶轨道电机车障碍物识别[J]. 浙江大学学报(工学版),2024,58(1):29-39. YANG Tun,GUO Yongcun,WANG Shuang,et al. Obstacle recognition of unmanned rail electric locomotive in underground coal mine[J]. Journal of Zhejiang University (Engineering Science),2024,58(1):29-39.
[3]	陈相蒙,王恩标,王刚. 煤矿电机车无人驾驶技术研究[J]. 煤炭科学技术,2020,48(增刊2):159-164. CHEN Xiangmeng,WANG Enbiao,WANG Gang. Research on electric locomotive self-driving technology in coal mine[J]. Coal Science and Technology,2020,48(S2):159-164.
[4]	韩江洪,卫星,陆阳,等. 煤矿井下机车无人驾驶系统关键技术[J]. 煤炭学报,2020,45(6):2104-2115. HAN Jianghong,WEI Xing,LU Yang,et al. Driverless technology of underground locomotive in coal mine[J]. Journal of China Coal Society,2020,45(6):2104-2115.
[5]	葛世荣. 煤矿机器人现状及发展方向[J]. 中国煤炭,2019,45(7):18-27. DOI: 10.3969/j.issn.1006-530X.2019.07.004 GE Shirong. Present situation and development direction of coal mine robots[J]. China Coal,2019,45(7):18-27. DOI: 10.3969/j.issn.1006-530X.2019.07.004
[6]	肖雨晴,杨慧敏. 目标检测算法在交通场景中应用综述[J]. 计算机工程与应用,2021,57(6):30-41. DOI: 10.3778/j.issn.1002-8331.2011-0361 XIAO Yuqing,YANG Huimin. Research on application of object detection algorithm in traffic scene[J]. Computer Engineering and Applications,2021,57(6):30-41. DOI: 10.3778/j.issn.1002-8331.2011-0361
[7]	REN Shaoqing,HE Kaiming,GIRSHICK R,et al. Faster R-CNN:towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(6):1137-1149. DOI: 10.1109/TPAMI.2016.2577031
[8]	GIRSHICK R,DONAHUE J,DARRELL T,et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]. IEEE Conference on Computer Vision and Pattern Recognition,Columbus,2014:580-587.
[9]	HE Kaiming,GKIOXARI G,DOLLÁR P,et al. Mask R-CNN[C]. IEEE International Conference on Computer Vision,Venice,2017:2980-2988.
[10]	LIU Wei,ANGUELOV D,ERHAN D,et al. SSD:single shot multibox detector[C]. European Conference on Computer Vision,Cham,2016:21-37.
[11]	REDMON J,FARHADI A. YOLO9000:better,faster,stronger[C]. IEEE Conference on Computer Vision and Pattern Recognition,Honolulu,2017:6517-6525.
[12]	REDMON J,FARHADI A. YOLOv3:an incremental improvement[EB/OL]. [2024-10-06]. https://arxiv.org/abs/1804.02767v1.
[13]	曹帅,董立红,邓凡,等. 基于YOLOv7−SE的煤矿井下场景小目标检测方法[J]. 工矿自动化,2024,50(3):35-41. CAO Shuai,DONG Lihong,DENG Fan,et al. A small object detection method for coal mine underground scene based on YOLOv7-SE[J]. Journal of Mine Automation,2024,50(3):35-41.
[14]	张传伟,张天乐,周李兵,等. 基于GCDB−YOLOv8的矿用无人车井下运输巷道工作人员检测方法[J/OL]. 煤炭学报:1-16[2025-04-15]. https://doi. org/10.13225/j. cnki. jccs. 2024.1298. ZHANG Chuanwei,ZHANG Tianle,ZHOU Libing,et al. Detection method of mining unmanned vehicles underground transportation roadway workers based on GCDB-YOLOv8[J/OL]. Journal of China Coal Society:1-16[2025-04-15]. https://doi.org/10.13225/j.cnki.jccs.2024.1298.
[15]	郑嘉祺. 基于DCNN的井下行人检测系统的研究与设计[D]. 西安:西安科技大学,2017. ZHENG Jiaqi. Research and design of underground pedestrian detection system based on DCNN[D]. Xi'an:Xi'an University of Science and Technology,2017.
[16]	李忠飞,冯仕咏,郭骏,等. 融合坐标注意力与多尺度特征的轻量级安全帽佩戴检测[J]. 工矿自动化,2023,49(11):151-159. LI Zhongfei,FENG Shiyong,GUO Jun,et al. Lightweight safety helmet wearing detection fusing coordinate attention and multiscale feature[J]. Journal of Mine Automation,2023,49(11):151-159.
[17]	陈伟,江志成,田子建,等. 基于YOLOv8的煤矿井下人员不安全动作检测算法[J]. 煤炭科学技术,2024,52(增刊2):267-283. CHEN Wei,JIANG Zhicheng,TIAN Zijian,et al. Unsafe action detection algorithm of underground personnel in coal mine based on YOLOv8[J]. Coal Science and Technology,2024,52(S2):267-283.
[18]	WANG Wenshan,WANG Shuang,GUO Yongcun,et al. Obstacle detection method of unmanned electric locomotive in coal mine based on YOLOv3-4L[J]. Journal of Electronic Imaging,2022,31(2). DOI: 10.1117/1.JEI.31.2.023032.
[19]	张楠楠,张晓,白铁成,等. 基于CBAM−YOLO v7的自然环境下棉叶病虫害识别方法[J]. 农业机械学报,2023,54(增刊1):239-244. ZHANG Nannan,ZHANG Xiao,BAI Tiecheng,et al. Identification method of cotton leaf pests and diseases in natural environment based on CBAM-YOLO v7[J]. Transactions of the Chinese Society for Agricultural Machinery,2023,54(S1):239-244.
[20]	李淇,石艳,范桃. 改进YOLOv8n的O型密封圈表面缺陷检测算法研究[J]. 计算机工程与应用,2024,60(18):126-135. DOI: 10.3778/j.issn.1002-8331.2405-0099 LI Qi,SHI Yan,FAN Tao. Research on O-ring surface defect detection algorithm based on improved YOLOv8n[J]. Computer Engineering and Applications,2024,60(18):126-135. DOI: 10.3778/j.issn.1002-8331.2405-0099
[21]	袁媛,白一超,周利东,等. 改进YOLOv8的输送带损伤检测方法[J/OL]. 中国机械工程:1-14[2025-04-22]. http://kns.cnki.net/kcms/detail/42.1294.th.20250421.1402.008.html. YUAN Yuan,BAI Yichao,ZHOU Lidong,et al. Conveyor belt damage detection based on improved YOLOv8[J/OL]. China Mechanical Engineering:1-14[2025-04-22]. http://kns.cnki.net/kcms/detail/42.1294.th.20250421.1402.008.html.
[22]	HU Jie,SHEN Li,SUN Gang. Squeeze-and-excitation networks[C]. IEEE/CVF Conference on Computer Vision and Pattern Recognition,Salt Lake City,2018:7132-7141.
[23]	TOUVRON H,CORD M,SABLAYROLLES A,et al. Going deeper with image transformers[C]. IEEE/CVF International Conference on Computer Vision,Montreal,2021:32-42.
[24]	OUYANG Daliang,HE Su,ZHANG Guozhong,et al. Efficient multi-scale attention module with cross-spatial learning[C]. IEEE International Conference on Acoustics,Speech and Signal Processing,Rhodes Island,2023:1-5.
[25]	PANDYA R V R. Generalized attention mechanism and relative position for transformer[EB/OL]. [2024-10-06]. https://arxiv.org/abs/2208.10247v1.

施引文献(8)

期刊类型引用(5)

1.	胡相捧，刘新华. 两柱掩护式液压支架初撑过程的机构演化机理. 煤炭科学技术. 2023(08): 239-249 . 百度学术
2.	金花，丁广炜，袁光英. 高架桥梁施工中梁体升降液压平衡自动控制方法. 自动化与仪器仪表. 2021(01): 115-118+123 . 百度学术
3.	迟振宇，王军辉，巩德华，贾晖，曹连民. 液压支架带压移架推移回路系统设计. 煤矿机械. 2021(04): 11-14 . 百度学术
4.	赵锐. 液压支架手动换向阀耐久性试验自动化装置设计. 工矿自动化. 2020(04): 104-108 . 本站查看
5.	李志旭，王庭明，孟祥强，李会，曹连民. 掩护式液压支架平衡控制回路补液系统设计分析. 煤矿机械. 2019(09): 30-33 . 百度学术