改进Yolov7算法的输送带损伤检测方法Improved Yolov7 Algorithm for Conveyor Belt Damage Detection
孙晖,胡志刚,田宇,王贡献
摘要(Abstract)：
实时可靠的输送带损伤检测对工业散货运输系统至关重要，现有的基于深度学习的视觉检测方法受输送带损伤样本量和特征丰富性不足等限制，导致其检测精度和泛化能力不足。针对上述问题，这里采用Yolov7视觉检测模型，提出了一种输送带损伤检测方法 Yolov7-SFL。首先，基于Yolov7模型构建输送带损伤特征提取与融合网络。然后，采用Sim-OTA策略实现了标签的精确分配，并提出了一种基于相似特征学习的联合训练策略来提升模型的特征学习能力、检测精度和泛化能力。最后，利用提出的双测试集DTS对所提方法进行试验验证。实验表明，这里提出的方法具有较高的检测速度、可靠性以及泛化能力，在双测试集DTS上，模型的mAP分别为97.57%和95.71%,FPS为30.59，满足输送带损伤实时检测要求。
关键词(KeyWords)： 输送带;损伤检测;标签分配;相似特征;联合训练
基金项目(Foundation): 科技创新2030—“新一代人工智能”重大项目（2022ZD0119304）
作者(Author): 孙晖,胡志刚,田宇,王贡献
DOI: 10.19356/j.cnki.1001-3997.20260403.011
参考文献(References)：
- [1]He D,Pang Y,Lodewijks G,et al.Healthy speed control of belt conveyors on conveying bulk materials[J].Powder Technology,2018(327):408-419.
- [2]武文.智能化选煤厂中防撕裂的应用[J].煤矿现代化，2017(3):23-25.(Wu Wen.The application of tear prevention in intelligent coal preparation plants[J].Modernization of Coal Mines,2017(3):23-25.)
- [3]郭启皇，乔铁柱.基于Otsu算法的输送带撕裂视觉检测系统研究[J].煤炭技术，2017,36(4):279-282.(Guo Qihuang,Qiao Tiezhu.Research on a visual detection system for conveyor belt tearing based on Otsu algorithm[J].Coal Technology,2017,36(4):279-282.)
- [4]肖书浩，吴蕾，何为，等.深度学习在表面质量检测方面的应用[J].机械设计与制造，2020(1):288-292.(Xiao Shuhao,Wu Lei,He Wei,et al.The application of deep learning in surface quality testing[J].Mechanical Design and Manufacturing,2020(1):288-292.)
- [5]布申申，田怀文.基于卷积神经网络的带钢表面缺陷检测算法[J].机械设计与制造，2022(7):29-33.(Bu Shenhen,Tian Huaiwen.Surface defect detection algorithm for strip steel based on convolutional neural network[J].Mechanical Design and Manufacturing,2022(7):29-33.)
- [6]Qu D R,Qiao T Z,Pang Y S,et al.Research on adcn method for damage detection of mining conveyor belt[J].Ieee Sensors Journal,2021,21(6):8662-8669.
- [7]Wang G X,Rao Z P,Sun H,et al.A belt tearing detection method of YOLOv4-BELT for multi-source interference environment[J]. Measurement,2022(189).
- [8]Wang G X,Liu Z Q,Sun H,et al.Yolox-BTFPN:An anchor-free conveyor belt damage detector with a biased feature extraction network[J].Measurement,2022(200).
- [9]Zhang S,Chi C,Yao Y,et al.Bridging the gap between anchor-based and anchor-free detection via adaptive training sample selection[C]. Proceedings of the IEEE/CVF Conference On Computer Vision and Pattern Recognition,2020:9759-9768.
- [10]Ma Y,Liu S,Li Z,et al.Iqdet:Instance-wise quality distribution sampling for object detection[C].Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,2021:1717-1725.
- [11]Ge Z,Liu S,Li Z,et al.Ota:Optimal transport assignment for object detection[C].Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,2021:303-312.
- [12]Wang C-Y,Bochkovskiy A,Liao H-Y M.YOLOv7:Trainable bag-offreebies sets new state-of-the-art for real-time object detectors[J].ArXiv Preprint ArXiv:2207.02696,2022.
- [13]Liu K,Sun Q,Sun D,et al.Underwater target detection based on improved YOLOv7[J].ArXiv Preprint ArXiv:2302.06939,2023.
- [14]Pham V,Nguyen D,Donan C.Road Damages Detection and Classification with YOLOv7[J]. arXiv preprint arXiv:2211.00091,2022.
- [15]Liu S,Qi L,Qin H F,et al.Path aggregation network for instance segmentation[J].2018 Ieee/Cvf Conference on Computer Vision and Pattern Recognition(Cvpr),2018:8759-8768.
- [16]Jain S,Salman H,Khaddaj A,et al.A data-based perspective on transfer learning[J].ArXiv Preprint ArXiv:2207.05739,2022.