一种基于Faster-RCNN的工业产品清点算法

doi:10.16731/j.cnki.1671-3133.2018.07.021

现代制造工程 ›› 2018, Vol. 454 ›› Issue (7): 124-128.doi: 10.16731/j.cnki.1671-3133.2018.07.021

• 仪器仪表/检测/监控 • 上一篇下一篇

一种基于Faster-RCNN的工业产品清点算法

朱泽民¹, 俞芳芳¹, 董蓉², 李勃¹

1 南京大学电子科学与工程学院,南京 210023;
2 南通大学电子信息学院,南通 226019

收稿日期:2017-03-27 出版日期:2018-07-20 发布日期:2018-07-20
通讯作者: 李勃,通信作者,博士,副教授,主要研究方向为视频图像处理、机器学习及模式识别。E-mail:liboee@nju.edu.cn
作者简介:朱泽民,硕士研究生,主要研究方向为机器视觉、图像处理。
基金资助:
国家自然科学基金项目(61401239)

Industrial product counting system based on Faster-RCNN

Zhu Zemin¹, Yu Fangfang¹, Dong Rong², Li Bo¹

1 School of Electronic Science and Engineering,Nanjing University,Nanjing 210023,China;
2 School of Electronic Information,Nantong University,Nantong 226019,Jiangsu,China

Received:2017-03-27 Online:2018-07-20 Published:2018-07-20

摘要/Abstract

摘要： 工业产品在出厂装箱的时候可能会出现漏装或混杂等异常情况,为了确保出厂质量,需要进行产品的清点。针对人工清点费时费力的问题,提出基于深度学习Faster-RCNN网络构造自动清点系统。针对遮挡情况下容易漏检的问题,创新地提出在训练阶段把产品特征碎片化,在检测阶段把碎片化的候选框的质心分别进行水平投影和垂直投影,并通过聚类确定产品的大致质心,最后将碎片化的候选框融合从而实现清点。实验证明,所提算法在提高检测效率的同时保证了非常高的可靠性和兼容性。

关键词: 产品清点, 深度学习, Faster-RCNN, 聚类, 碎片化

Abstract: There may be missing workpiece and other defects when circular workpieces are packed.To ensure the quality,need to check the number of workpieces in the box.A new auto counting system based on Faster-RCNN is proposed to save labor cost.To handle the problem that the workpiece may be undetected when be sheltered by something,a newly fragmentation training method is proposed.After that,it merges the sub-proposal by horizontal projection and vertical projection and clustering the centroid of the sub-proposals when testing.Realistic experiments prove that the method improves detecting efficiency as well as ensuring very high reliability and compatibility.

Key words: workpiece counting, deep learning, Faster-RCNN, clustering, fragmentation

中图分类号:

TP391

朱泽民, 俞芳芳, 董蓉, 李勃. 一种基于Faster-RCNN的工业产品清点算法[J]. 现代制造工程, 2018, 454(7): 124-128.

Zhu Zemin, Yu Fangfang, Dong Rong, Li Bo. Industrial product counting system based on Faster-RCNN[J]. Modern Manufacturing Engineering, 2018, 454(7): 124-128.

参考文献

[1] 李华伟,李凤婷,余天洪,等.一种用于缺陷检测的瓶盖图像外轮廓确定方法:CN103606167A[P].2014.
[2] JIA Y,SHELHAMER E,DONAHUE J,et al.Caffe:Convolutional Architecture for Fast FeatureEmbedding.ACM International Conference on Multimedia[C].[S.l.]:ACM,2014:675-678.
[3] REN S,HE K,GIRSHICK R,et al.Faster R-CNN:Towards real-time object detection with region proposal networks.Advances in neural information processing systems[C].[S.l.]:[s.n.],2015:91-99.
[4] SALVADOR A,GIRO-I-Nieto X,MARQUES F,et al.Faster R-CNN Features for Instance Search.Deepvision Workshop at CVPR[C].[S.l.]:[s.n.],2016.
[5] JIANG H, LEARNED-MILLER E.Face detection with the faster R-CNN.Automatic Face & Gesture Recognition (FG 2017).2017 12th IEEE International Conference on[C].[S.l.]:IEEE,2017:650-657.
[6] HOANG Ngan Le T,ZHENG Y,ZHU C,et al.Multiple Scale Faster-RCNN Approach to Driver's Cell-Phone Usage and Hands on Steering Wheel Detection.Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops[C].[S.l.]:[s.n.],2016:46-53.
[7] LI J, LIANG X, SHEN S M, et al. Scale-aware Fast R-CNN for Pedestrian Detection[J]. IEEE Transactions on Multimedia, 2015(99):1.
[8] CAI Z,FAN Q,FERIS R S,et al.A unified multi-scale deep convolutional neural network for fast object detection.European Conference on Computer Vision.Springer International Publishing[C].[S.l.]:[s.n.],2016:354-370.
[9] ZENG X,OUYANG W,YANG B,et al.Gated bi-directional cnn for object detection.European Conference on Computer Vision.Springer International Publishing[C].[S.l.]:[s.n.]2016:354-369.
[10] CHUA L O.CNN: A paradigm for complexity[M].[S.l.]:World Scientific, 1998.
[11] GIRSHICK R.Fast R-CNN.IEEE International Conference on Computer Vision[C].[S.l.]:IEEE,2015:1440-1448.
[12] ZEILER M D,FERGUS R.Visualizing and understanding convolutional networks.European Conference on Computer Vision[C].[S.l.]:Springer International Publishing,2014:818-833.
[13] 吴再龙,张云泉,徐建良,等.基于OpenCL的Kmeans算法的优化研究[J].计算机科学与探索,2014,8(10):1162-1176.
[14] WAGSTAFF K, CARDIE C, ROGERS S. Constrained K-means Clustering with Background Knowledge. Eighteenth International Conference on Machine Learning[C].[S.l.]:Morgan Kaufmann Publishers Inc.,2001:577-584.
[15] OWENS J D,HOUSTON M,LUEBKE D,et al.GPU computing[J].Proceedings of the IEEE,2008,96(5):879-899.

一种基于Faster-RCNN的工业产品清点算法

Industrial product counting system based on Faster-RCNN

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

[1]	马宇超，付华良，吴鹏，陈信华，王鼎，陈帅，曹晨雨. 深度网络自适应优化的Mask R-CNN模型在铸件表面缺陷检测中的应用研究[J]. 现代制造工程, 2022, 499(4): 112-118.
[2]	韩婷，石宇强. 基于振动信号融合的ACO-DCNN多工况设备故障诊断[J]. 现代制造工程, 2021, 492(9): 94-100.
[3]	李亚平，李素杰，马波，郭俊霞. 基于TCN的滚动轴承振动趋势与剩余寿命预测研究[J]. 现代制造工程, 2021, 492(9): 124-131.
[4]	肖乾浩. 基于机器学习理论的机械故障诊断方法综述[J]. 现代制造工程, 2021, 490(7): 148-161.
[5]	伍萍辉，陈新，张馨，段丹丹，唐冰霞，张云鹤. 基于聚类-改进灰狼算法的设施番茄分割识别方法[J]. 现代制造工程, 2021, 489(6): 83-89.
[6]	陈国才. 基于函数型大数据聚类分析的机械故障在线检测[J]. 现代制造工程, 2021, 489(6): 142-147.
[7]	盛永健，黄子龙，刘晨，曹毅，张洪. 基于改进卷积神经网络的燃气调压器故障识别研究[J]. 现代制造工程, 2021, 487(4): 132-138.
[8]	王峰，张亮亮，王海波，陶小亮，王学宇. 基于K-means聚类算法的电驱产线监测模型的建立与软件实现[J]. 现代制造工程, 2021, 485(2): 127-133.
[9]	陈登凯，王憬鸾，郭家言，李方舟，李志. 基于功能元参数相关性的产品功能模块设计方法研究[J]. 现代制造工程, 2020, 478(7): 94-101.
[10]	金颖，王学影，段林茂. 基于LabVIEW和Mask R-CNN的柱塞式制动主缸内槽表面缺陷检测[J]. 现代制造工程, 2020, 476(5): 125-132.
[11]	贾海利，张健，祁宇明，邓三鹏. 云制造平台资源需求的高效匹配策略研究[J]. 现代制造工程, 2020, 475(4): 36-43.
[12]	魏弦. 基于有序聚类的机床热误差测点优化[J]. 现代制造工程, 2018, 451(4): 108-114.
[13]	沈振辉, 杨拴强. 基于模糊聚类及相关性分析的温度测点布置优化方法研究[J]. 现代制造工程, 2018, 458(11): 112-118.
[14]	李翔, 李珂, 李忠学. 一种基于模糊C均值算法确定制造单元数的方法[J]. 现代制造工程, 2017, 440(5): 132-138.
[15]	汪金鹏, 郭阳宽, 祝连庆, 潘志康. 基于聚类的关节式坐标测量机姿态优化方法[J]. 现代制造工程, 2017, 438(3): 141-144.