改进卷积神经网络算法在机械零件实时识别与定位中的应用
|
摘要
通用的目标识别与定位卷积神经网络算法难以兼顾精度和速度的要求。本文在YOLO v2卷积神经网络的基础上,采用多尺度训练、网络预训练和k-means维度聚类等优化方法,提出了机械零件实时识别与定位的改进卷积神经网络算法。本文以螺母和垫片2种物体为识别与定位的对象,以工业传送带为场景,同时考虑到了传送带上干扰物的存在,对改进算法的准确率和速度进行了实验测试。实验结果证明本文的算法相对其它常用目标检测卷积神经网络算法在识别准确率和速度上达到了很好的平衡,为零件实时分拣提供了基础。
Universal target recognition and location of convolutional neural network algorithms are difficult to balance the accuracy and speed requirements. Based on YOLO v2 convolutional neural network,this paper adopts multi-scale training,network pre-training and k-means dimension clustering optimization methods to propose an improved convolutional neural network algorithm for real-time recognition and location of mechanical parts. In this paper,two kinds of objects,nuts and pads,are used to identify and locate the object. The industrial conveyor belt is taken as the scene. At the same time,the existence of interferences on the conveyor belt is considered. The accuracy and speed of the improved algorithm are tested experimentally. Compared with other common target detection convolutional neural network algorithms,the algorithm achieves a good balance in the recognition accuracy and speed,and provides a basis for real-time sorting of parts.
Universal target recognition and location of convolutional neural network algorithms are difficult to balance the accuracy and speed requirements. Based on YOLO v2 convolutional neural network,this paper adopts multi-scale training,network pre-training and k-means dimension clustering optimization methods to propose an improved convolutional neural network algorithm for real-time recognition and location of mechanical parts. In this paper,two kinds of objects,nuts and pads,are used to identify and locate the object. The industrial conveyor belt is taken as the scene. At the same time,the existence of interferences on the conveyor belt is considered. The accuracy and speed of the improved algorithm are tested experimentally. Compared with other common target detection convolutional neural network algorithms,the algorithm achieves a good balance in the recognition accuracy and speed,and provides a basis for real-time sorting of parts.
引文
[1]张丛.基于机器视觉的机械零件的识别研究[D].天津:天津职业技术师范大学,2015.
[2]管经纬.基于机器视觉的工件自动分拣系统研究[D].上海:东华大学,2015.
[3]黄咨.基于卷积神经网络的目标检测模型的研究[D].上海:上海交通大学,2014.
[4] KRIZHEVSKY A,SUTSKEVER I,HINTON G E. Imagenet classification w ith deep convolutional neural netw orks[C]//Advances in neural information processing systems. Lake Tahoe,Nevada,USA:dblp,2012:1097-1105.
[5] SZEGEDY C, LIU W, JIA Y, et al. Going deeper with convolutions[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Boston,M A,USA:IEEE,2015:1-9.
[6] GIRSHICK R,DONAHUE J,DARELL T,et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.Columbus,Ohio:IEEE,2014:580-587.
[7] GIRSHICK R. Fast r-cnn[C]//Proceedings of the IEEE International Conference on Computer Vision. Santiago,Chile:IEEE,2015:1440-1448.
[8]REN Shaoqing,HE Kaiming,GIRSHICK R,et al. Faster RCNN:Tow ards real-time object detection w ith region proposal netw orks[J]. IEEE Transactions on Pattern Analysis and M achine Intelligence,2017,39(6):1137-1149.
[9] REDMON J,DIWALA S,GIRSHICK R,et al. You only look once:Unified,real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Las Vegas,Nevada:IEEE,2016:779-788.
[10]REDMON J,FARHADI A. YOLO9000:Better,faster,stronger[J]. ar Xiv preprint ar Xiv:1612.08242,2016.
[11]郭丽丽,丁世飞.深度学习研究进展[J].计算机科学,2015,42(5):28-33.
[12]LECUN Y,BOTTOU L,BENGIO Y,et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE,1998,86(11):2278-2324.
[13]IOFFES,SZEGEDY C. Batch normalization:Accelerating deep netw ork training by reducing internal covariate shift[J]. ar Xiv preprint ar Xiv:1502.03167,2015.
[14]SIMONYANK,ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[J]. ar Xiv preprint ar Xiv:1409.1556,2014.
[15]LIU Wei,ANGUELOV D,ERHAN D,et al. SSD:Single shot multibox detector[M]//LEIBE B,M ATAS J,SEBE N,ed.Computer Vision-ECCV 2016. ECCV 2016. Lecture Notes in Computer Science,Cham:Springer,2016,9905:21-37.
[16]LINM,CHEN Q,YAN S. Network in network[J]. ar Xiv preprint ar Xiv:1312.4400,2013.
[17]魏湧明,全吉成,侯宇青阳.基于YOLO v2的无人机航拍图像定位研究[J].激光与光电子学进展,2017,54(11):101-110.
[18]张素洁,赵怀慈.最优聚类个数和初始聚类中心点选取算法研究[J].计算机应用研究,2017,34(6):1617-1620.
[2]管经纬.基于机器视觉的工件自动分拣系统研究[D].上海:东华大学,2015.
[3]黄咨.基于卷积神经网络的目标检测模型的研究[D].上海:上海交通大学,2014.
[4] KRIZHEVSKY A,SUTSKEVER I,HINTON G E. Imagenet classification w ith deep convolutional neural netw orks[C]//Advances in neural information processing systems. Lake Tahoe,Nevada,USA:dblp,2012:1097-1105.
[5] SZEGEDY C, LIU W, JIA Y, et al. Going deeper with convolutions[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Boston,M A,USA:IEEE,2015:1-9.
[6] GIRSHICK R,DONAHUE J,DARELL T,et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.Columbus,Ohio:IEEE,2014:580-587.
[7] GIRSHICK R. Fast r-cnn[C]//Proceedings of the IEEE International Conference on Computer Vision. Santiago,Chile:IEEE,2015:1440-1448.
[8]REN Shaoqing,HE Kaiming,GIRSHICK R,et al. Faster RCNN:Tow ards real-time object detection w ith region proposal netw orks[J]. IEEE Transactions on Pattern Analysis and M achine Intelligence,2017,39(6):1137-1149.
[9] REDMON J,DIWALA S,GIRSHICK R,et al. You only look once:Unified,real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Las Vegas,Nevada:IEEE,2016:779-788.
[10]REDMON J,FARHADI A. YOLO9000:Better,faster,stronger[J]. ar Xiv preprint ar Xiv:1612.08242,2016.
[11]郭丽丽,丁世飞.深度学习研究进展[J].计算机科学,2015,42(5):28-33.
[12]LECUN Y,BOTTOU L,BENGIO Y,et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE,1998,86(11):2278-2324.
[13]IOFFES,SZEGEDY C. Batch normalization:Accelerating deep netw ork training by reducing internal covariate shift[J]. ar Xiv preprint ar Xiv:1502.03167,2015.
[14]SIMONYANK,ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[J]. ar Xiv preprint ar Xiv:1409.1556,2014.
[15]LIU Wei,ANGUELOV D,ERHAN D,et al. SSD:Single shot multibox detector[M]//LEIBE B,M ATAS J,SEBE N,ed.Computer Vision-ECCV 2016. ECCV 2016. Lecture Notes in Computer Science,Cham:Springer,2016,9905:21-37.
[16]LINM,CHEN Q,YAN S. Network in network[J]. ar Xiv preprint ar Xiv:1312.4400,2013.
[17]魏湧明,全吉成,侯宇青阳.基于YOLO v2的无人机航拍图像定位研究[J].激光与光电子学进展,2017,54(11):101-110.
[18]张素洁,赵怀慈.最优聚类个数和初始聚类中心点选取算法研究[J].计算机应用研究,2017,34(6):1617-1620.