基于深度学习的型钢建材智能计数与在线部署
陈宇韬 , 陈隽 , 陈文豪 , 黄茜 , 李洋 , 刘思媛
建筑钢结构进展 ›› 2025, Vol. 27 ›› Issue (06) : 104 -113.
基于深度学习的型钢建材智能计数与在线部署
Intelligent Counting and Online Deployment of Steel Building Materials Based on Deep Learning
建材管控是建筑智能施工建造的重要一环,由于型钢截面形状多样、摆放时相互堆叠遮挡且堆放背景环境复杂等因素,其智能实时计数的研究与应用相对于钢筋、钢管等圆形截面建材仍充满挑战。对此,本文提出了一种基于改进YOLOv7的非圆形截面建材的计数模型。首先拍摄大量工字钢、方钢管和木枋的现场照片,结合数据增强手段获得14 950张图像,共包含1 290 210个端面。进而在现有YOLOv7架构中新增一阶段实例分割模块,与目标检测任务并行操作,再通过改进骨干网络和检测头、加入注意力机制、修改损失函数和训练策略等提升模型检测精度,最后采用轻量级卷积模块减少特征冗余以实现模型轻量化并提升检测时效。试验结果表明,三种经过改进的建材计数模型的检测准确率均达到90%以上,满足使用要求。最后,将上述算法集成到小程序并已成功部署,使用户通过手机拍照即可完成型钢的在线实时计数工作。
Building materials control is an important part of building intelligent construction, where section steel pose challenges compared with rebar and steel pipe due to its diverse cross-sectional shapes, mutual stacking and occlusion of end faces, and complex background environments in intelligent real-time counting. This study proposes a counting model for non-circular cross-section building materials based on improved YOLOv7. First, a large number of field photographs of I-beams, square steel tubes, and wooden keels were taken and augmented to 14,950 photographs, with a total of 1,290,210 end faces, using data enhancement. Then, a one-stage instance segmentation module was added to the YOLOv7 network to achieve instance segmentation and target detection tasks in parallel. Furthermore, the original YOLOv7 network was enhanced by refining the backbone network and detection head, introducing an attention mechanism, modifying the loss function, and implementing training strategy to improve the model's detection accuracy. Finally, a lightweight convolution module was used to reduce feature redundancy, enabling effective compression of the model and improving the detection time. The results indicate that the detection accuracy of all three improved building material counting models exceeds 90%, meeting the usage requirements. The above algorithm was integrated into an applet, which has been successfully deployed so that users can complete the real-time counting work by taking photos with their mobile phones.
目标检测 / 目标分割 / 施工建材 / 非圆形截面 / 实时计数 / 模型检测精度
target detection / instance segmentation / building material / non-circular section / real-time counting / detection accuracy of model
| [1] |
廖玉平. 加快建筑业转型 推动高质量发展——解读《关于推动智能建造与建筑工业化协同发展的指导意见》[J]. 中国勘察设计,2020(9):20-21. |
| [2] |
LIAO Yuping. Accelerating the transformation of construction industry and promoting high-quality development:Interpretation of guiding opinions on promoting the coordinated development of intelligent construction and building industrialization[J]. China Engineering Consulting,2020(9):20-21. (in Chinese) |
| [3] |
王亦君,蒋首超. 基于计算机视觉的钢构件防腐涂层缺陷检测[J]. 建筑钢结构进展,2023,25(12):85-93,101. DOI:10.13969/j.cnki.cn31-1893.2023.12.009. |
| [4] |
WANG Yijun,JIANG Shouchao. Defect identification of anti-corrosive coating of steel members based on computer vision[J]. Progress in Steel Building Structures,2023,25(12):85-93,101. DOI:10.13969/j.cnki.cn31-1893. 2023.12.009.(in Chinese) |
| [5] |
RAN B D,LI S X,ZHOU B C,et al. Deep learning based detection and counting algorithm for rebar images[C]//2023 35th Chinese Control and Decision Conference (CCDC). New York:IEEE,2023. |
| [6] |
XIN Y,XUE W,YUAN P X,et al. Research on an automatic counting method for steel bars' image[C]//2010 International Conference on Electrical and Control Engineering. New York:IEEE,2010. |
| [7] |
WU Y,ZHOU X F,ZHANG Y C. Steel bars counting and splitting method based on machine vision[C]//2015 IEEE International Conference on Cyber Technology in Automation,Control,and Intelligent Systems (CYBER). New York:IEEE,2015. |
| [8] |
GHAZALI M F,WONG L K,SEE J. Automatic detection and counting of circular and rectangular steel bars[C]//9th International Conference on Robotic,Vision,Signal Processing and Power Applications. Singapore:Springer Singapore,2016. |
| [9] |
WANG H,POLDEN J,JIRGENS J,et al. Automatic rebar counting using image processing and machine learning[C]//2019 IEEE 9th Annual International Conference on CYBER Technology in Automation,Control,and Intelligent Systems (CYBER). New York:IEEE,2019. |
| [10] |
FAN Z,LU J W,QIU B Z,et al. Automated steel bar counting and center localization with convolutional neural networks[EB/OL]. [2019-06-19]. https://arxiv.org/abs/1906.00891v2. |
| [11] |
LI Y,LU Y J,CHEN J. A deep learning approach for real-time rebar counting on the construction site based on YOLOv3 detector[J]. Automation in Construction,2021,124:103602. DOI:10.1016/j.autcon.2021.103602. |
| [12] |
LI Y,CHEN J. Computer vision–based counting model for dense steel pipe on construction sites[J]. Journal of Construction Engineering and Management,2022,148(1):04021178. DOI:10.1061/(ASCE)CO.1943-7862.0002217. |
| [13] |
刘浩,辛山. 面向密集型钢筋计数的GCA-MobilenetV2-YOLOv4算法[J]. 电子测量技术,2023,46(9):166-174. DOI:10.19651/j.cnki.emt.2210228. |
| [14] |
LIU Hao,XIN Shan. GCA-MobilenetV2-YOLOv4 algorithm for intensive rebar counting[J]. Electronic Measurement Technology,2023,46(9):166-174. DOI:10.19651/j.cnki.emt.2210228.(in Chinese) |
| [15] |
方成,于盛鑫,李永刚, 基于深度学习的土木工程计算机视觉健康监测[J]. 同济大学学报(自然科学版),2024,52(2):213-222. DOI:10.11908/j.issn.0253-374x.22370. |
| [16] |
FANG Cheng,YU Shengxin,LI Yonggang,et al. Deep learning-based computer vision for health monitoring in civil engineering[J]. Journal of Tongji University (Natural Science),2024,52(2):213-222. DOI:10.11908/j.issn.0253-374x.22370.(in Chinese) |
| [17] |
陈隽,陈文豪,李洋. 基于目标检测的施工钢材物料智能实时计数[J]. 同济大学学报(自然科学版),2023,51(11):1701-1710. DOI:10.11908/j.issn.0253-374x.22157. |
| [18] |
CHEN Jun,CHEN Wenhao,LI Yang. Intelligent real-time counting of construction materials based on object detection[J]. Journal of Tongji University (Natural Science),2023,51(11):1701-1710. DOI:10.11908/j.issn.0253-374x.22157.(in Chinese) |
| [19] |
WANG C Y,BOCHKOVSKIY A,LIAO H M. YOLOv7:Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]//2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). New York:IEEE,2023. |
| [20] |
WOO S,PARK J,LEE J Y,et al. CBAM:Convolutional block attention module[C]//Computer Vision-ECCV 2018. Berlin:Springer,2018. |
| [21] |
MA N N,ZHANG X Y,SUN J. Funnel activation for visual recognition[C]//Computer Vision-ECCV 2020. Berlin:Springer,2020. |
| [22] |
GEVORGYAN Z. SIoU loss:More powerful learning for bounding box regression[EB/OL]. [2022-05-25]. https://arxiv.org/abs/2205.12740v1. |
| [23] |
HAN K,WANG Y H,TIAN Q,et al. GhostNet:More features from cheap operations[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). New York:IEEE,2020. |
| [24] |
BOLYA D,ZHOU C,XIAO F Y,et al. YOLACT:Real-time instance segmentation[C]//2019 IEEE/CVF International Conference on Computer Vision (ICCV). New York:IEEE,2019. |
| [25] |
ZHENG Z H,WANG P,LIU W,et al. Distance-IoU loss:Faster and better learning for bounding box regression[J]. Proceedings of the AAAI Conference on Artificial Intelligence,2020,34(7):12993-13000. DOI:10.1609/aaai.v34i07.6999. |
| [26] |
CHATTOPADHAY A,SARKAR A,HOWLADER P,et al. Grad-CAM:Generalized gradient-based visual explanations for deep convolutional networks[C]//2018 IEEE Winter Conference on Applications of Computer Vision (WACV). New York:IEEE,2018. |
同济大学2022年度学科交叉联合攻关项目(2022-3-YB-06)
/
| 〈 |
|
〉 |