Coal gangue detection based on CBA-YOLO model
-
摘要: 煤矸石检测中存在样本间特征差异小、目标密集等问题，导致现有煤矸石检测方法精度不高且实时性较差。针对该问题，提出了一种基于CBA−YOLO模型的煤矸石检测方法。CBA−YOLO模型以速度较快、精度较高的YOLOv5m为基础模型，在YOLOv5m的Backbone中加入卷积块注意力模块（CBAM），通过串联空间注意力模块和通道注意力模块，在聚焦特征差异的同时降低数据维度，提高煤矸石检测性能；在Neck部分采用双向特征金字塔网络（BiFPN）结构，通过融合不同尺度的特征提高模型计算效率，从而提升煤矸石检测速度；在Prediction部分采用Alpha−IoU函数作为损失函数，通过设置权重系数加速对高置信度目标的学习，进一步提高煤矸石检测精度。实验结果表明：CBA−YOLO模型对煤矸石的平均检测精度达98.2%，比YOLOv5模型提高了3.4%，检测速度提升了10%；CBA−YOLO模型的鲁棒性更强，可有效避免漏检、误检和重叠现象。Abstract: There are some problems in coal gangue detection, such as small differences of characteristics between samples and dense targets. This leads to low precision and poor real-time performance of the existing coal gangue detection methods. In order to solve this problem, a method of coal gangue detection based on CBA-YOLO model is proposed. The CBA-YOLO model is based on YOLOv5m, which has faster speed and higher precision. The convolutional block attention module (CBAM) is added to the Backbone of YOLOv5m. The spatial attention module and the channel attention module are connected in series to focus on the difference of characteristics and reduce the data dimension. And the detection performance of coal gangue is improved. In the Neck part, the bi-directional feature pyramid network (BiFPN) structure is adopted to improve the calculation efficiency of the model by integrating the features of different scales. Therefore, the detection speed of coal gangue is improved. In the Prediction part, the Alpha-IoU function is used as the loss function. And the weight coefficient is set to accelerate the learning of high confidence targets, so as to further improve the detection precision of coal gangue. The experimental results show that the average detection precision of CBA-YOLO model for coal gangue is 98.2%, which is 3.4% higher than that of YOLOv5 model. The detection speed is increased by 10%. CBA-YOLO model is more robust and can effectively avoid missed detection, false detection and overlap.
-
图 5 基于CBA−YOLO模型的煤矸石检测流程
Figure 5. Flow of coal gangue detection based on CBA-YOLO model
表 1 消融实验结果
Table 1. Results of ablation experiment
模型 mAP/% 帧率/（帧·s−1） 模型 mAP/% 帧率/（帧·s−1） YOLOv5 94.8 30 YOLO−CB 96.2 35.2 YOLO−C 95.7 32.2 YOLO−CA 97.4 30.5 YOLO−B 95.5 34.1 YOLO−BA 97.5 32 YOLO−A 96.0 29.8 CBA−YOLO 98.2 33 
下载: 导出CSV
-
[1] 张振红. 我国干法选煤技术发展现状与应用前景[J]. 选煤技术,2019(1):43-47,52.ZHANG Zhenhong. China's coal dry cleaning technology−state-of-the-art and application prospect[J]. Coal Preparation Technology,2019(1):43-47,52. [2] SU Lingling, CAO Xiangang, MA Hongwei, et al. Research on coal gangue identification by using convolutional neural network[C]//IEEE Advanced Information Management, Communicates, Electronic and Automation Control Conference, Xi'an, 2018: 810-814. [3] HONG Huichao, ZHENG Lixin, ZHU Jianqing, et al. Automatic recognition of coal and gangue based on convolution neural network[EB/OL]. （2017-12-03）[2022-01-05]. https://arxiv.org/abs/1712.00720. [4] 赵明辉. 一种煤矸石优化识别方法[J]. 工矿自动化,2020,46(7):113-116.ZHAO Minghui. A coal-gangue optimization identification method[J]. Industry and Mine Automation,2020,46(7):113-116. [5] GIRSHICK R. Fast R-CNN[C]//IEEE International Conference on Computer Vision, Chile, 2015. [6] LIU Wei, ANGUELOV D, ERHAN D, et al. SSD: single shot multibox detector[C]// European Conference on Computer Vision, 2016: 21-37. [7] REDMON J, FARHADI A. Yolov3: an incremental improvement[EB/OL]. （2018-04-08） [2022-01-05]. https://arxiv.org/abs/1804.02767. [8] 沈科,季亮,张袁浩,等. 基于改进YOLOv5s模型的煤矸目标检测[J]. 工矿自动化,2021,47(11):107-111,118.SHEN Ke,JI Liang,ZHANG Yuanhao,et al. Research on coal and gangue detection algorithm based on improved YOLOv5s model[J]. Industry and Mine Automation,2021,47(11):107-111,118. [9] REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection[C]//IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, 2016: 779-788. [10] BOCHKOVSKIY A, WANG C Y, LIAO H Y M. YOLOv4: optimal speed and accuracy of object detection[EB/OL]. （2020-04-23） [2022-01-05]. https://arxiv.org/abs/2004.10934. [11] DENG Jun,XUAN Xiaojing,WANG Weifeng,et al. A review of research on object detection based on deep learning[J]. Journal of Physics Conference Series,2020,1684:012028. doi: 10.1088/1742-6596/1684/1/012028 [12] HU Jie,SHEN Li,ALBANIE S,et al. Squeeze-and-excitation networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2020,42(8):2011-2023. doi: 10.1109/TPAMI.2019.2913372 [13] LI Xiang, WANG Wenhai, HU Xiaolin, et al. Selective kernel networks[EB/OL]. （2019-05-18）[2022-01-06]. https://arxiv.org/abs/1903.06586. [14] WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[EB/OL]. （2018-07-18）[2022-01-06]. https://arxiv.org/abs/1807.06521. [15] TAN M, PANG R, LE Q V. Efficientdet: scalable and efficient object detection[EB/OL]. （2020-07-27）[2022-01-06]. https://arxiv.org/abs/1911.09070. [16] JIANG Borui, LUO Ruixuan, MAO Jiayuan, et al. Acquisition of localization confidence for accurate object detection[C]// European Conference on Computer Vision, 2018: 816-832. [17] REZATOFIGHI H, TSOI N, GWAK J Y, et al. Generalized intersection over union: a metric and a loss for bounding box regression [EB/OL]. （2019-04-15）[2022-01-06]. https://arxiv.org/abs/1902.09630. [18] ZHENG Zhaohui, WANG Ping, REN Dongwei, et al. Enhancing geometric factors in model learning and inference for object detection and instance segmentation[EB/OL]. （2021-07-05）[2022-01-06]. https://arxiv.org/abs/2005.03572. [19] HE Jiabo, ERFANI S, MA Xingjun, et al. Alpha-IoU: a family of power intersection over union losses for bounding box regression[EB/OL]. （2021-10-26）[2022-01-06]. https://arxiv.org/abs/2110.13675. -
点击查看大图
图(9) / 表(1)
计量
- 文章访问数: 381
- HTML全文浏览量: 154
- PDF下载量: 68
- 被引次数: 0
