基于YOLOv5s−FSW模型的选煤厂煤矸检测研究

燕碧娟; 王凯民; 郭鹏程; 郑馨旭; 董浩; 刘勇

doi:10.13272/j.issn.1671-251x.2023100090

基于YOLOv5s−FSW模型的选煤厂煤矸检测研究

doi: 10.13272/j.issn.1671-251x.2023100090

1.
太原科技大学机械工程学院，山西太原　030024
2.
山西人工智能矿山创新实验室有限公司，山西太原　030032

基金项目: 山西省重点研发计划项目（202102010101010）。

详细信息

作者简介:
燕碧娟（1975—），女，山西芮城人，教授，博士，主要研究方向为智能矿山视觉感知关键技术，E-mail：tyustybj@tyust.edu.cn

中图分类号: TD948.9
计量
- 文章访问数: 145
- HTML全文浏览量: 41
- PDF下载量: 30
- 被引次数: 0
出版历程
- 收稿日期: 2023-10-27
- 修回日期: 2024-05-06
- 网络出版日期: 2024-06-13

Research on coal gangue detection in coal preparation plant based on YOLOv5s-FSW model

1.
College of Mechanical Engineering, Taiyuan University of Science and Technology, Taiyuan 030024, China
2.
Shanxi Artificial Intelligence Mine Innovation Laboratory Co., Ltd., Taiyuan 030032, China

摘要

摘要: 针对现有煤矸检测模型存在的特征提取不充分、参数量大、检测精度低且实时性差等问题，提出了一种基于YOLOv5s−FSW模型的选煤厂煤矸检测方法。该模型在YOLOv5s的基础上进行改进，首先将主干网络的C3模块替换为FasterNet Block结构，通过降低模型的参数量和计算量提高检测速度；然后，在颈部网络引入无参型SimAM注意力机制，增强模型对复杂环境下重要目标的关注，进一步提高模型的特征提取能力；最后，在输出端用Wise−IoU替换CIoU边界框损失函数，使模型聚焦普通质量锚框，提高收敛速度和边框的检测精度。消融实验结果表明：与YOLOv5s模型相比，YOLOv5s−FSW模型的平均精度均值（mAP）提高了1.9%，模型权重减少了0.6 MiB，参数量减少了4.7%，检测速度提高了19.3%。对比实验结果表明：YOLOv5s−FSW模型的mAP达95.8%，较YOLOv5s−CBC，YOLOv5s−ASA，YOLOv5s−SDE模型分别提高了1.1%，1.5%和1.2%，较YOLOv5m，YOLOv6s模型分别提高了0.3%，0.6%；检测速度达36.4帧/s，较YOLOv5s−CBC，YOLOv5s−ASA模型分别提高了28.2%和20.5%，较YOLOv5m，YOLOv6s，YOLOv7模型分别提高了16.3%，15.2%，45.0%。热力图可视化实验结果表明：YOLOv5s−FSW模型对煤矸目标特征区域更加敏感且关注度更高。检测实验结果表明：在环境昏暗、图像模糊、目标相互遮挡的复杂场景下，YOLOv5s−FSW模型对煤矸目标检测的置信度得分高于YOLOv5s模型，且有效避免了误检和漏检现象的发生。
- 煤矸检测 /
- YOLOv5s /
- FasterNet Block /
- SimAM注意力机制 /
- Wise−IoU边界框损失函数
Abstract: A coal gangue detection method in coal preparation plant based on YOLOv5s-FSW model is proposed to address the problems of insufficient feature extraction, large parameter quantity, low detection precision, and poor real-time performance in existing coal gangue detection models. This model is improved on the basis of YOLOv5s. Firstly, the C3 module in the Backbone section is replaced with a FasterNet Block structure, which improves detection speed by reducing the number of model parameters and computation. Secondly, in the Neck section, a parameter free SimAM attention mechanism is introduced to enhance the model's attention to important targets in complex environments, further improving the model's feature extraction capability. Finally, in the Prediction layer, the CIoU bounding box loss function is replaced with Wise-IoU, and the model focuses on ordinary quality anchor boxes to improve convergence speed and bounding box detection precision. The results of the ablation experiment indicate that compared with the YOLOv5s model, The mean average precision (mAP) of the YOLOv5s-FSW model has been improved by 1.9%, the model weight has been reduced by 0.6 MiB, the number of parameters has been reduced by 4.7%, and the detection speed has been improved by 19.3%. The comparative experimental results show that the YOLOv5s-FSW model has a mAP of 95.8%, which is 1.1%, 1.5%, and 1.2% higher compared to the YOLOv5s-CBC, YOLOv5s-ASA, and YOLOv5s-SDE models, respectively, and compared to YOLOv5m, YOLOv6s improved by 0.3%, 0.6% respectively. The detection speed of the YOLOv5s-FSW reaches 36.4 frames per second, which is 28.2% and 20.5% higher than the YOLOv5s-CBC and YOLOv5s-ASA models, respectively. Compared to YOLOv5m, YOLOv6s and YOLOv7, the detection speed of the YOLOv5s-FSW has increased by 16.3%, 15.2%, and 45.0%, respectively. The visualization experiment results of the thermal map show that the YOLOv5s-FSW model is more sensitive to the target feature areas of coal gangue and has higher attention. The detection experiment results show that in complex scenes with dim environments, blurred images, and mutual occlusion of targets, the YOLOv5s-FSW model has a higher confidence score for coal gangue target detection than the YOLOv5s model, and effectively avoids the occurrence of false positives and missed detection.
- coal gangue detection /
- YOLOv5s /
- FasterNet Block /
- SimAM attention mechanism /
- Wise IoU bounding box loss function

HTML全文

图 1 YOLOv5s−FSW网络结构

Figure 1. YOLOv5s-FSW network structure

下载: 全尺寸图片幻灯片

图 2 FasterNet Block结构

Figure 2. FasterNet Block structure

下载: 全尺寸图片幻灯片

图 3 SimAM注意力机制

Figure 3. SimAM attention mechanism

下载: 全尺寸图片幻灯片

图 4 数据集示例

Figure 4. Dataset example

下载: 全尺寸图片幻灯片

图 5 模型改进前后煤矸热力图结果对比

Figure 5. Comparison of thermal map results of coal-gangue before and after model improvement

下载: 全尺寸图片幻灯片

图 6 模型改进前后煤矸检测效果对比

Figure 6. Comparison of coal-gangue detection results before and after model improvement

下载: 全尺寸图片幻灯片

表 1 消融实验结果

Table 1. Ablation experiment results

模型	精确率/%	召回率/%	mAP/%	权重/MiB	计算量	参数量/10⁵	检测速度/（帧·s⁻¹）
YOLOv5s	89.9	88.6	93.9	13.7	15.8	70.2	30.5
改进模型1	89.6	85.1	93.5	13.1	14.3	66.9	37.8
改进模型2	89.9	89.1	94.2	13.7	15.8	70.2	29.1
改进模型3	91.1	89.7	95.3	13.7	15.8	70.2	28.3
YOLOv5s−FSW	91.8	90.1	95.8	13.1	14.3	66.9	36.4

下载: 导出CSV

表 2 不同检测模型性能对比

Table 2. Performance comparison of different detection models

模型	mAP/%	权重/MiB	计算量	检测速度/（帧·s⁻¹）
YOLOv5s−CBC	94.7	15.3	15.9	28.4
YOLOv5s−ASA	94.3	13.4	15.6	30.2
YOLOv5s−SDE	94.6	12.7	12.1	37.8
YOLOv5s	93.9	13.7	15.8	30.5
YOLOv5m	95.5	40.2	47.9	31.3
YOLOv6s	95.2	38.7	45.2	31.6
YOLOv7	96.1	71.3	105.2	25.1
YOLOv5s−FSW	95.8	13.1	14.3	36.4

下载: 导出CSV

参考文献(27)

[1]	金智新,曹孟涛,王宏伟. “中等收入”与新“双控”背景下煤炭行业转型发展新机遇[J]. 煤炭科学技术,2023,51(1):45-58. JIN Zhixin,CAO Mengtao,WANG Hongwei. New opportunities for coal industry transformation and development under the background of the level of a moderately developed country and a new "dual control" system[J]. Coal Science and Technology,2023,51(1):45-58.
[2]	李君清,李寅琪. 煤炭产业经济走势及煤炭企业对策研究[J]. 中国煤炭,2023,49(3):16-22. doi: 10.3969/j.issn.1006-530X.2023.03.003 LI Junqing,LI Yinqi. Study on the development trend of coal industry economy and countermeasures of coal enterprises[J]. China Coal,2023,49(3):16-22. doi: 10.3969/j.issn.1006-530X.2023.03.003
[3]	周宏春. 新型能源体系破解能源保供与降碳双重压力研究与探讨[J]. 中国煤炭,2023,49(5):1-10. doi: 10.3969/j.issn.1006-530X.2023.05.001 ZHOU Hongchun. Research and discussion on breaking the dual pressure of energy supply guarantee and carbon reduction by the new energy system[J]. China Coal,2023,49(5):1-10. doi: 10.3969/j.issn.1006-530X.2023.05.001
[4]	朱吉茂,孙宝东,张军,等. “双碳”目标下我国煤炭资源开发布局研究[J]. 中国煤炭,2023,49(1):44-50. doi: 10.3969/j.issn.1006-530X.2023.01.006 ZHU Jimao,SUN Baodong,ZHANG Jun,et al. Research on China's coal resources development layout under the goals of carbon peak and carbon neutrality[J]. China Coal,2023,49(1):44-50. doi: 10.3969/j.issn.1006-530X.2023.01.006
[5]	唐珏,王俊. “双碳”目标下煤炭发展及对策建议[J]. 中国矿业,2023,32(9):22-31. doi: 10.12075/j.issn.1004-4051.20230483 TANG Jue,WANG Jun. Coal development and countermeasures under the carbon peaking and carbon neutrality goals[J]. China Mining Magazine,2023,32(9):22-31. doi: 10.12075/j.issn.1004-4051.20230483
[6]	郭静,李磊,李志明. 干法选煤技术创新进展及其节能节水降污效果分析[J]. 中国煤炭,2022,48(5):68-75. doi: 10.3969/j.issn.1006-530X.2022.05.012 GUO Jing,LI Lei,LI Zhiming. Innovation progress of dry coal preparation technology and analysis of its effect of energy saving,water saving and pollution reduction[J]. China Coal,2022,48(5):68-75. doi: 10.3969/j.issn.1006-530X.2022.05.012
[7]	刘志杰. 重介洗煤技术在选煤厂的应用[J]. 能源与节能,2023(7):136-138. doi: 10.3969/j.issn.2095-0802.2023.07.036 LIU Zhijie. Application of heavy medium coal washing technology in coal preparation plant[J]. Energy and Energy Conservation,2023(7):136-138. doi: 10.3969/j.issn.2095-0802.2023.07.036
[8]	ZHANG Ningbo,LIU Changyou. Radiation characteristics of natural gamma-ray from coal and gangue for recognition in top coal caving[J]. Scientific Reports,2018,8(1):190. doi: 10.1038/s41598-017-18625-y
[9]	韩子彬,王丽宏,申志刚,等. 基于X射线分选方法在选煤厂中的应用[J]. 煤炭科学技术,2022,50(增刊1):327-332. HAN Zibin,WANG Lihong,SHEN Zhigang,et al. Application of X-ray separation method in coal preparation plant[J]. Coal Science and Technology,2022,50(S1):327-332.
[10]	蔡秀凡,谢金辰. YOLOv4煤矸石检测方法研究[J]. 煤炭工程,2022,54(8):157-162. CAI Xiufan,XIE Jinchen. YOLOv4-based detection method of coal and gangue[J]. Coal Engineering,2022,54(8):157-162.
[11]	来文豪,周孟然,胡锋,等. 基于多光谱成像和改进YOLOv4的煤矸石检测[J]. 光学学报,2020,40(24):72-80. LAI Wenhao,ZHOU Mengran,HU Feng,et al. Coal gangue detection based on multi-spectral imaging and improved YOLOv4[J]. Acta Optica Sinica,2020,40(24):72-80.
[12]	高如新,常嘉浩,杜亚博,等. 基于改进YOLOv5s的煤矸石目标检测算法[J]. 电子测量技术,2023,46(13):95-101. GAO Ruxin,CHANG Jiahao,DU Yabo,et al. Coal gangue target detection algorithm based on improved YOLOv5s[J]. Electronic Measurement Technology,2023,46(13):95-101.
[13]	郑道能. 一种改进的tiny YOLOv3煤矸石快速识别模型[J]. 工矿自动化,2023,49(4):113-119. ZHENG Daoneng. An improved tiny YOLOv3 rapid recognition model for coal-gangue[J]. Journal of Mine Automation,2023,49(4):113-119.
[14]	陈彪,卢兆林,代伟,等. 基于轻量化HPG−YOLOX−S模型的煤矸石图像精准识别[J]. 工矿自动化,2022,48(11):33-38. CHEN Biao,LU Zhaolin,DAI Wei,et al. Accurate recognition of coal-gangue image based on lightweight HPG-YOLOX-S model[J]. Journal of Mine Automation,2022,48(11):33-38.
[15]	桂方俊,李尧. 基于CBA−YOLO模型的煤矸石检测[J]. 工矿自动化,2022,48(6):128-133. GUI Fangjun,LI Yao. Coal gangue detection based on CBA-YOLO model[J]. Journal of Mine Automation,2022,48(6):128-133.
[16]	张释如,黄综浏,张袁浩,等. 基于改进YOLOv5的煤矸识别研究[J]. 工矿自动化,2022,48(11):39-44. ZHANG Shiru,HUANG Zongliu,ZHANG Yuanhao,et al. Coal and gangue recognition research based on improved YOLOv5[J]. Journal of Mine Automation,2022,48(11):39-44.
[17]	张磊,王浩盛,雷伟强,等. 基于YOLOv5s−SDE的带式输送机煤矸目标检测[J]. 工矿自动化,2023,49(4):106-112. ZHANG Lei,WANG Haosheng,LEI Weiqiang,et al. Coal gangue target detection of belt conveyor based on YOLOv5s-SDE[J]. Journal of Mine Automation,2023,49(4):106-112.
[18]	REDMON J,FARHADI A. YOLOv3:an incremental improvement[C]. IEEE Conference on Computer Vision and Pattern Recognition,Salt Lake City,2018:89-95.
[19]	芦碧波,周允,李小军,等. 融合注意力机制的YOLOv5轻量化煤矿井下人员检测算法[J]. 煤炭技术,2023,42(10):200-203. LU Bibo,ZHOU Yun,LI Xiaojun,et al. YOLOv5 lightweight coal mine underground personnel detection algorithm base on attention mechanism[J]. Coal Technology,2023,42(10):200-203.
[20]	CHEN Jierun,KAO Shiuhong,HE Hao,et al. Run,don't walk:chasing higher FLOPS for faster neural networks[C]. IEEE/CVF Conference on Computer Vision and Pattern Recognition,Vancouver,2023:12021-12031.
[21]	HU Jie,SHEN Li,SUN Gang. Squeeze-and-excitation networks[C]. IEEE Conference on Computer Vision and Pattern Recognition,Salt Lake City,2018:7132-7141.
[22]	WOO S,PARK J C,LEE J Y,et al. Cbam:convolutional block attention module[C]. European Conference on Computer Vision,Munich,2018:3-19.
[23]	柏罗,张宏立,王聪. 基于高效注意力和上下文感知的目标跟踪算法[J]. 北京航空航天大学学报,2022,48(7):1222-1232. BAI Luo,ZHANG Hongli,WANG Cong. Target tracking algorithm based on efficient attention and context awareness[J]. Journal of Beijing University of Aeronautics and Astronautics,2022,48(7):1222-1232.
[24]	YANG Lingxiao,ZHANG Ruyuan,LI Lida,et al. Simam:a simple,parameter-free attention module for convolutional neural networks[C]. International Conference on Machine Learning,New York,2021:11863-11874.
[25]	JIANG Borui,LUO Ruixuan,MAO Jiayuan,et al. Acquisition of localization confidence for accurate object detection[C]. European Conference on Computer Vision,Munich,2018:816-832.
[26]	ZHENG Zhaohui,WANG Ping,REN Dongwei,et al. Enhancing geometric factors in model learning and inference for object detection and instance segmentation[J]. IEEE Transactions on Cybernetics,2021,52(8):8574-8586.
[27]	TONG Zanjia,CHEN Yuhang,XU Zewei,et al. Wise−IoU:bounding box regression loss with dynamic focusing mechanism[J]. Computer Science,2023. DOI: 10.48550/arXiv.2301.10051.