基于Faster R-CNN的办公用品目标检测
|
摘要
RCNN网络与全卷积网络框架等技术使得目标检测技术能够快速发展。RCNN网络与全卷积网络框架不仅训练速度快,推断速度也十分的迅速,此外还具有良好的鲁棒性以及灵活性。在人工智能领域的发展中,提高目标检测效率的关键在于好的技术,以及得到更加有效的、深层的特征表示,通过使用深层网络的多层结构来简洁地表达复杂函数。本文用到的目标检测方法先要用区域建议网络得到建议位置再进行检测,因为Fast R-CNN和R-CNN等目标检测算法已经在运行时间方面有了很大的提高,所以计算区域建议成为目标检测的一个计算瓶颈。本文通过在算法中加入特征融合技术,将每一卷积层提取的特征进行融合,使用区域建议网络来进行候选区域提取。区域建议网络和检测网络共享全图的卷积特征,从而很大程度地缩短候选区域的提取时间,提高目标检测的精度。
The technologies of RCNN network and full convolution network framework enable target detection technology to develop rapidly.RCNN networks and full convolution network frameworks are not only fast in training,but also very fast in inference speed.Besides,it has good robustness and flexibility.In the development of artificial intelligence,the key to improve the efficiency of target detection lies in good technology.We need to get a more effective and deep feature representation,which can express complex functions simply by using the multi-layer structure of deep network.The target detection method used in this paper is first to use the regional recommended network to get the proposed location and then test,because the Fast R-CNN and R-CNN target detection algorithms have been greatly improved in the running time,so the calculation area is suggested to be a computing bottle neck of the target detection.This paper joins the feature fusion technology in the algorithm,combines the features extracted from each layer of the accumulated layer,uses the regional recommended network to extract the candidate region.The regional recommendation network and the detection network share the convolution feature of the whole graph,so that the extraction time of the candidate region can be greatly shortened and the accuracy of target detection is improved.
The technologies of RCNN network and full convolution network framework enable target detection technology to develop rapidly.RCNN networks and full convolution network frameworks are not only fast in training,but also very fast in inference speed.Besides,it has good robustness and flexibility.In the development of artificial intelligence,the key to improve the efficiency of target detection lies in good technology.We need to get a more effective and deep feature representation,which can express complex functions simply by using the multi-layer structure of deep network.The target detection method used in this paper is first to use the regional recommended network to get the proposed location and then test,because the Fast R-CNN and R-CNN target detection algorithms have been greatly improved in the running time,so the calculation area is suggested to be a computing bottle neck of the target detection.This paper joins the feature fusion technology in the algorithm,combines the features extracted from each layer of the accumulated layer,uses the regional recommended network to extract the candidate region.The regional recommendation network and the detection network share the convolution feature of the whole graph,so that the extraction time of the candidate region can be greatly shortened and the accuracy of target detection is improved.
引文
[1]CHAVALI N,AGRAWAL H,MAHENDRU A,et al.Object-proposal evaluation protocol is'gameable'[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition.2016:835-844.
[2]丁小可.基于深度学习的目标检测算法研究[J/OL].机电信息,2018(33):154-155[2018-11-25].http://www.cnki.com.cn/Article/CJFDTotal-JDXX2018112202G.htm.
[3]DAI J F,HE K M,SUN J.Convolutional feature masking for joint object and stuff segmentation[C]//Proceedings of2015 IEEE Conference on Computer Vision and Pattern Recognition.2015:3992-4000.
[4]REN S Q,HE K M,GIRSHICK R,et al.Faster R-CNN:Towards real-time object detection with region proposal networks[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(6):1137-1149.
[5]HOSANG J,BENENSON R,Dollár P,et al.What makes for effective detection proposals?[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2016,38(4):814-830.
[6]PINHEIRO P O,COLLOBERT R,Dollár P.Learning to segment object candidates[C]//Proceedings of the 28th International Conference on Neural Information Processing Systems.2015:1990-1998.
[7]LIU W,ANGUELOV D,ERHAN D,et al.Ssd:Single shot multibox detector[C]//European Conference on Computer Vision.2016:21-37.
[8]REDMON J,DIVVALA S,GIRSHICK R,et al.You only look once:Unified,real-time object detection[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition.2016:779-788.
[9]LONG J,SHELHAMER E,DARRELL T.Fully convolutional networks for semantic segmentation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(4):640-651.
[10]UIJLINGS J R R,SANDE K E A V D,GEVERS T,et al.Selective search for object recognition[J].International Journal of Computer Vision,2013,104(2):154-171.
[11]HARIHARAN B,ARBELEz P,GIRSHICK R,et al.Simultaneous detection and segmentation[C]//European Conference on Computer Vision.2014:297-312.
[12]ERHAN D,SZEGEDY C,TOSHEV A,et al.Scalable object detection using deep neural networks[C]//Proceedings of 2014 IEEE Conference on Computer Vision and Pattern Recognition.2014:2147-2154.
[13]KRIZHEVSKY A,SUTSKEVER I,HINTON G E.Image Net classification with deep convolutional neural networks[C]//Proceedings of the 25th International Conference on Neural Information Processing Systems.2012:1097-1105.
[14]曹靖康,段江永,孟娟.基于深度特征融合的空间植物图像分割算法[J].计算机与现代化,2018(10):58-62.
[15]DAI J F,LI Y,HE K M,et al.R-FCN:Object detection via region-based fully convolutional networks[J].Computer Vision and Pattern Recognition,arxiv:1605.06409,2016.
[16]CHENG M M,WARRELL J,LIN W Y,et al.Efficient salient region detection with soft image abstraction[C]//2013 IEEE International Conference on Computer Vision.2013:1529-1536.
[17]GIRSHICK R.Fast R-CNN[J].Computer Science,arxiv:1504.08083,2015.
[18]HE K M,ZHANG X Y,REN S Q,et al.Spatial pyramid pooling in deep convolutional networks for visual recognition[C]//European Conference on Computer Vision-EC-CV 2014.2014:346-361.
[19]郭少军,沈同圣,徐健,等.基于Object Ness BING的海面多舰船目标检测[J].系统工程与电子技术,2016,38(1):14-20.
[20]RANTALANKILA P,KANNALA J,RAHTU E.Generating object segmentation proposals using global and local search[C]//2014 IEEE Conference on Computer Vision and Pattern Recognition.2014:2417-2424.
[21]ENDRES I,HOIEM D.Category-independent object proposals with diverse ranking[J].IEEE Transactions on Pattern A-nalysis and Machine Intelligence,2014,36(2):222-234.
[22]ZHANG Z M,WARRELL J,TORR P H S.Proposal generation for object detection using cascaded ranking SVMs[C]//2011 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).2011:1497-1504.
[23]GANIN Y,LEMPITSKY V.N4-fields:Neural network nearest neighbor fields for image transforms[C]//Asian Conference on Computer Vision.2014:536-551.
[24]RAHTU E,KANNALA J,BLASCHKO M.Learning a category independent object detection cascade[C]//2011IEEE International Conference on Computer Vision(IC-CV).2011:1052-1059.
[25]EVERINGHAM M,GOOL L V,WILLIAMS C K I,et al.The Pascal visual object classes(VOC)challenge[J].International Journal of Computer Vision,2010,88(2):303-338.
[2]丁小可.基于深度学习的目标检测算法研究[J/OL].机电信息,2018(33):154-155[2018-11-25].http://www.cnki.com.cn/Article/CJFDTotal-JDXX2018112202G.htm.
[3]DAI J F,HE K M,SUN J.Convolutional feature masking for joint object and stuff segmentation[C]//Proceedings of2015 IEEE Conference on Computer Vision and Pattern Recognition.2015:3992-4000.
[4]REN S Q,HE K M,GIRSHICK R,et al.Faster R-CNN:Towards real-time object detection with region proposal networks[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(6):1137-1149.
[5]HOSANG J,BENENSON R,Dollár P,et al.What makes for effective detection proposals?[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2016,38(4):814-830.
[6]PINHEIRO P O,COLLOBERT R,Dollár P.Learning to segment object candidates[C]//Proceedings of the 28th International Conference on Neural Information Processing Systems.2015:1990-1998.
[7]LIU W,ANGUELOV D,ERHAN D,et al.Ssd:Single shot multibox detector[C]//European Conference on Computer Vision.2016:21-37.
[8]REDMON J,DIVVALA S,GIRSHICK R,et al.You only look once:Unified,real-time object detection[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition.2016:779-788.
[9]LONG J,SHELHAMER E,DARRELL T.Fully convolutional networks for semantic segmentation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(4):640-651.
[10]UIJLINGS J R R,SANDE K E A V D,GEVERS T,et al.Selective search for object recognition[J].International Journal of Computer Vision,2013,104(2):154-171.
[11]HARIHARAN B,ARBELEz P,GIRSHICK R,et al.Simultaneous detection and segmentation[C]//European Conference on Computer Vision.2014:297-312.
[12]ERHAN D,SZEGEDY C,TOSHEV A,et al.Scalable object detection using deep neural networks[C]//Proceedings of 2014 IEEE Conference on Computer Vision and Pattern Recognition.2014:2147-2154.
[13]KRIZHEVSKY A,SUTSKEVER I,HINTON G E.Image Net classification with deep convolutional neural networks[C]//Proceedings of the 25th International Conference on Neural Information Processing Systems.2012:1097-1105.
[14]曹靖康,段江永,孟娟.基于深度特征融合的空间植物图像分割算法[J].计算机与现代化,2018(10):58-62.
[15]DAI J F,LI Y,HE K M,et al.R-FCN:Object detection via region-based fully convolutional networks[J].Computer Vision and Pattern Recognition,arxiv:1605.06409,2016.
[16]CHENG M M,WARRELL J,LIN W Y,et al.Efficient salient region detection with soft image abstraction[C]//2013 IEEE International Conference on Computer Vision.2013:1529-1536.
[17]GIRSHICK R.Fast R-CNN[J].Computer Science,arxiv:1504.08083,2015.
[18]HE K M,ZHANG X Y,REN S Q,et al.Spatial pyramid pooling in deep convolutional networks for visual recognition[C]//European Conference on Computer Vision-EC-CV 2014.2014:346-361.
[19]郭少军,沈同圣,徐健,等.基于Object Ness BING的海面多舰船目标检测[J].系统工程与电子技术,2016,38(1):14-20.
[20]RANTALANKILA P,KANNALA J,RAHTU E.Generating object segmentation proposals using global and local search[C]//2014 IEEE Conference on Computer Vision and Pattern Recognition.2014:2417-2424.
[21]ENDRES I,HOIEM D.Category-independent object proposals with diverse ranking[J].IEEE Transactions on Pattern A-nalysis and Machine Intelligence,2014,36(2):222-234.
[22]ZHANG Z M,WARRELL J,TORR P H S.Proposal generation for object detection using cascaded ranking SVMs[C]//2011 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).2011:1497-1504.
[23]GANIN Y,LEMPITSKY V.N4-fields:Neural network nearest neighbor fields for image transforms[C]//Asian Conference on Computer Vision.2014:536-551.
[24]RAHTU E,KANNALA J,BLASCHKO M.Learning a category independent object detection cascade[C]//2011IEEE International Conference on Computer Vision(IC-CV).2011:1052-1059.
[25]EVERINGHAM M,GOOL L V,WILLIAMS C K I,et al.The Pascal visual object classes(VOC)challenge[J].International Journal of Computer Vision,2010,88(2):303-338.