指向性边缘框:基于内法向特征的一般物体检测

doi:10.1007/s11390-017-1752-9

指向性边缘框:基于内法向特征的一般物体检测

1 School of Electronic Information and Communications, Huazhong University of Science and Technology Wuhan 430074, China;
2 Key Laboratory of Specialty Fiber Optics and Optical Access Networks, Shanghai University, Shanghai 200444, China

基金项目: This work was supported in part by the National Natural Science Foundation of China under Grant Nos. 61573160 and 61672336, and in part by the "Chen Guang" Project supported by Shanghai Municipal Education Commission and Shanghai Education Development Foundation of China under Grant No. 15CG43.

计量
- 文章访问数: 41
- HTML全文浏览量: 0
- PDF下载量: 806
出版历程
- 收稿日期: 2016-12-15
- 修回日期: 2017-05-14
- 发布日期: 2017-07-04

Directional Edge Boxes: Exploiting Inner Normal Direction Cues for Effective Object Proposal Generation

1 School of Electronic Information and Communications, Huazhong University of Science and Technology Wuhan 430074, China;
2 Key Laboratory of Specialty Fiber Optics and Optical Access Networks, Shanghai University, Shanghai 200444, China

Funds: This work was supported in part by the National Natural Science Foundation of China under Grant Nos. 61573160 and 61672336, and in part by the "Chen Guang" Project supported by Shanghai Municipal Education Commission and Shanghai Education Development Foundation of China under Grant No. 15CG43.

More Information

Corresponding author:
Wei Shen E-mail: wei.shen@t.shu.edu.cn

摘要

摘要: 边缘是一种定位物体候选框的重要特征。最近在这个问题上的进展大多数是通过定义基于边缘特征的有效物体置性度来推动。在本文中，我们提出一种新的叫做指向性边缘的表示。所谓指向性边缘，是指该边缘上每个点给予了一个指向物体中心的方法。该方向通过卷积神经网络学习得到。基于指向性边缘，我们提出两种新的物体置性度用于对物体候选框排序。在PASCAL VOC 2007测试数据集上，产生1000个候选框，我们的方法可以分别取得97.1%和81.9%召回率，在IoU阈值为分别为0.5和0.7的情况下。该实验结果表明，我们的方法性能优于当前最好的方法。
- 物体候选框 /
- 指向性边缘 /
- 卷积神经网络
Abstract: Edges are important cues for localizing object proposals. The recent progresses to this problem are mostly driven by defining effective objectness measures based on edge cues. In this paper, we develop a new representation named directional edges on which each edge pixel is assigned with a direction toward object center, through learning a direction prediction model with convolutional neural networks in a holistic manner. Based on directional edges, two new objectness measures are designed for ranking object proposals. Experiments show that the proposed method achieves 97.1% object recall at an overlap threshold of 0.5 and 81.9% object recall at an overlap threshold of 0.7 at 1 000 proposals on the PASCAL VOC 2007 test dataset, which is superior to the state-of-the-art methods.
- object proposal /
- directional edge /
- convolutional neural network

HTML全文

参考文献()

[1]	Lim J, Zitnick C, Dollár P. Sketch tokens:A learned midlevel representation for contour and object detection. In Proc. CVPR, June 2013, pp.3158-3165.
[2]	Dollár P, Zitnick C. Structured forests for fast edge detection. In Proc. ICCV, December 2013, pp.1841-1848.
[3]	Bertasius G, Shi J, Torresani L. DeepEdge:A multi-scale bifurcated deep network for top-down contour detection. In Proc. CVPR, June 2015, pp.4380-4389.
[4]	Shen W, Wang X, Wang Y, Bai X, Zhang Z. DeepContour:A deep convolutional feature learned by positivesharing loss for contour detection. In Proc. CVPR, June 2015, pp.3982-3991.
[5]	Xie S, Tu Z. Holistically-nested edge detection. In Proc. ICCV, December 2015, pp.1395-1403.
[6]	Zitnick C L, Dollár P. Edge Boxes:Locating object proposals from edges. In Proc. ECCV, September 2014, pp.391-405.
[7]	Cheng M M, Zhang Z, Lin W Y, Torr P. BING:Binarized normed gradients for objectness estimation at 300fps. In Proc. CVPR, June 2014, pp.3286-3293.
[8]	Lu C, Liu S, Jia J, Tang C K. Contour Box:Rejecting object proposals without explicit closed contours. In Proc. ICCV, December 2015, pp.2021-2029.
[9]	Qi Y, Song Y Z, Xiang T, Zhang H, Hospedales T, Li Y, Guo J. Making better use of edges via perceptual grouping. In Proc. CVPR, June 2015, pp.1856-1865.
[10]	Long J, Shelhamer E, Darrell T. Fully convolutional networks for semantic segmentation. In Proc. CVPR, June 2015, pp.3431-3440.
[11]	Xia G S, Delon J, Gousseau Y. Shape-based invariant texture indexing. International Journal of Computer Vision, 2010, 88(3):382-403.
[12]	Xia G S, Delon J, Gousseau Y. Accurate junction detection and characterization in natural images. International Journal of Computer Vision, 2014, 106(1):31-56.
[13]	Xie J, Dai G, Zhu F, Wong E, Fang Y. DeepShape:Deeplearned shape descriptor for 3D shape retrieval. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(7):1335-1345.
[14]	Dai G, Xie J, Zhu F, Fang Y. Learning a discriminative deformation-invariant 3D shape descriptor via many-to-one encoder. Pattern Recognition Letters, 2016, 83:330-338.
[15]	Everingham M, van Gool L, Williams C K, Winn J, Zisserman A. The Pascal visual object classes (VOC) challenge. IJCV, 2010, 88(2):303-338.
[16]	Alexe B, Deselaers T, Ferrari V. What is an object? In Proc. CVPR, June 2010, pp.73-80.
[17]	van de Sande K E, Uijlings J R, Gevers T, Smeulders A W. Segmentation as selective search for object recognition. In Proc. ICCV, November 2011, pp.1879-1886.
[18]	Yanulevskaya V, Uijlings J, Sebe N. Learning to group objects. In Proc. CVPR, June 2014, pp.3134-3141.
[19]	Manen S, Guillaumin M, van Gool L. Prime object proposals with Randomized Prim's algorithm. In Proc. ICCV, December 2013, pp.2536-2543.
[20]	Xiao Y, Lu C, Tsougenis E, Lu Y, Tang C K. Complexityadaptive distance metric for object proposals generation. In Proc. CVPR, June 2015, pp.778-786.
[21]	Rantalankila P, Kannala J, Rahtu E. Generating object segmentation proposals using global and local search. In Proc. CVPR, June 2014, pp.2417-2424.
[22]	Arbeláez P, Pont-Tuset J, Barron J, Marques F, Malik J. Multiscale combinatorial grouping. In Proc. CVPR, June 2014, pp.328-335.
[23]	Endres I, Hoiem D. Category-independent object proposals with diverse ranking. IEEE Trans. PAMI, 2014, 36(2):222-234.
[24]	Humayun A, Li F, Rehg J. RIGOR:Reusing inference in graph cuts for generating object regions. In Proc. CVPR, June 2014, pp.336-343.
[25]	Humayun A, Li F, Rehg J M. The middle child problem:Revisiting parametric mincut and seeds for object proposals. In Proc. ICCV, December 2015, pp.1600-1608.
[26]	Krähenbühl P, Koltun V. Geodesic object proposals. In Proc. ECCV, Sept. 2014, pp.725-739.
[27]	Lee T, Fidler S, Dickinson S. Learning to combine midlevel cues for object proposal generation. In Proc. ICCV, December 2015, pp.1680-1688.
[28]	Wang C, Zhao L, Liang S, Zhang L, Jia J, Wei Y. Object proposal by multibranch hierarchical segmentation. In Proc. CVPR, June 2015, pp.3873-3881.
[29]	Pinheiro P O, Collobert R, Dollár P. Learning to segment object candidates. In Proc. Advances in Neural Information Processing Systems, Dec. 2015, pp.1990-1998.
[30]	Martin D R, Fowlkes C C, Malik J. Learning to detect natural image boundaries using local brightness, color, and texture cues. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2004, 26(5):530-549.
[31]	Arbelaez P, Maire M, Fowlkes C, Malik J. Contour detection and hierarchical image segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(5):898-916.
[32]	Ren X, Bo L. Discriminatively trained sparse code gradients for contour detection. In Proc. Advances in Neural Information Processing Systems, December 2012, pp.593-601.
[33]	Hwang J J, Liu T L. Pixel-wise deep learning for contour detection. arXiv:1504.01989, 2015. https://arxiv.org/abs-/1504.01989, May 2017.
[34]	Rahtu E, Kannala J, Blaschko M. Learning a category independent object detection cascade. In Proc. ICCV, November 2011, pp.1052-1059.
[35]	Kuo W, Hariharan B, Malik J. DeepBox:Learning objectness with convolutional networks. In Proc. ICCV, December 2015, pp.2479-2487.
[36]	Ghodrati A, Diba A, Pedersoli M, Tuytelaars T, van Gool L. DeepProposal:Hunting objects by cascading deep convolutional layers. In Proc. CVPR, June 2015, pp.2578-2586.
[37]	Chen X, Ma H, Wang X, Zhao Z. Improving object proposals with multithresholding straddling expansion. In Proc. CVPR, June 2015, pp.2587-2595.
[38]	Zhang Z, Liu Y, Bolukbasi T, Cheng M M, Saligrama V. BING++:A fast high quality object proposal generator at 100fps. arXiv:1511.04511, 2015. https://arxiv.org/abs/-1511.04511, Apr. 2017.
[39]	Xiao Y,Wu J, Yuan J. mCENTRIST:A multichannel feature generation mechanism for scene categorization. IEEE Trans. Image Processing, 2014, 23(2):823-836.
[40]	Fang Z, Cao Z, Xiao Y, Zhu L, Yuan J. Adobe Boxes:Locating object proposals using object adobes. IEEE Trans. Image Processing, 2016, 25(9):4116-4128.
[41]	He S, Lau R W. Oriented object proposals. In Proc. the IEEE International Conference on Computer Vision, Dec. 2015, pp.280-288.
[42]	Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition. arXiv:1409.1556, 2014. https://arxiv.org/abs/1409.1556, May 2017.
[43]	Lin T Y, Maire M, Belongie S, Hays J, Perona P, Ramanan D, Dollár P, Zitnick C L. Microsoft COCO:Common objects in context. In Proc. the 13th European Conference on Computer Vision, September 2014, pp.740-755.
[44]	Alexe B, Deselaers T, Ferrari V. Measuring the objectness of image windows. IEEE Trans. PAMI, 2012, 34(11):2189-2202.
[45]	Zhang Z, Warrell J, Torr P H. Proposal generation for object detection using cascaded ranking SVMs. In Proc. CVPR, June 2011, pp.1497-1504.
[46]	Krähenbühl P, Koltun V. Learning to propose objects. In Proc. CVPR, June 2015, pp.1574-1582.
[47]	Ballard D H. Generalizing the Hough transform to detect arbitrary shapes. Pattern Recognition, 1981, 13(2):111-122.
[48]	Borgefors G. Hierarchical chamfer matching:A parametric edge matching algorithm. IEEE Trans. PAMI, 1988, 10(6):849-865.

施引文献

资源附件()

点击查看大图

计量

文章访问数: 41
HTML全文浏览量: 0
PDF下载量: 806
被引次数: 0

指向性边缘框:基于内法向特征的一般物体检测

计量

Directional Edge Boxes: Exploiting Inner Normal Direction Cues for Effective Object Proposal Generation

Corresponding author:
Wei Shen E-mail: wei.shen@t.shu.edu.cn

计量

目录

Home

Overview

Resources

Contents

指向性边缘框:基于内法向特征的一般物体检测

计量

出版历程

Directional Edge Boxes: Exploiting Inner Normal Direction Cues for Effective Object Proposal Generation

Corresponding author: Wei Shen E-mail: wei.shen@t.shu.edu.cn

计量

出版历程

目录

Home

Overview

Resources

Contents

Corresponding author:
Wei Shen E-mail: wei.shen@t.shu.edu.cn