Learning detectors quickly using structured covariance matrices

Computer vision is increasingly becoming interested in the rapid estimation of object detectors. Canonical hard negative mining strategies are slow as they require multiple passes of the large negative training set. Recent work has demonstrated that if the distribution of negative examples is assumed to be stationary, then Linear Discriminant Analysis (LDA) can learn comparable detectors without ever revisiting the negative set. Even with this insight, however, the time to learn a single object detector can still be on the order of tens of seconds on a modern desktop computer. This paper proposes to leverage the resulting structured covariance matrix to obtain detectors with identical performance in orders of magnitude less time and memory. We elucidate an important connection to the correlation filter literature, demonstrating that these can also be trained without ever revisiting the negative set.

[1]  W. F. Trench An Algorithm for the Inversion of Finite Toeplitz Matrices , 1964 .

[2]  R. Chan Circulant preconditioners for Hermitian Toeplitz systems , 2011 .

[3]  Rui Caseiro,et al.  Beyond Hard Negative Mining: Efficient Detector Learning via Block-Circulant Decomposition , 2013, 2013 IEEE International Conference on Computer Vision.

[4]  Keinosuke Fukunaga,et al.  Introduction to Statistical Pattern Recognition , 1972 .

[5]  Tony F. Chan,et al.  Circulant preconditioners for Toeplitz-block matrices , 1994, Numerical Algorithms.

[6]  B. V. K. Vijaya Kumar,et al.  Maximum Margin Correlation Filter: A New Approach for Localization and Classification , 2013, IEEE Transactions on Image Processing.

[7]  Jitendra Malik,et al.  Poselets: Body part detectors trained using 3D human pose annotations , 2009, 2009 IEEE 12th International Conference on Computer Vision.

[8]  Fei-Fei Li,et al.  ImageNet: A large-scale hierarchical image database , 2009, 2009 IEEE Conference on Computer Vision and Pattern Recognition.

[9]  Justin K. Romberg,et al.  Image deconvolution via superfast inversion of a class of two-level Toeplitz matrices , 2012, 2012 19th IEEE International Conference on Image Processing.

[10]  G. Strang A proposal for toeplitz matrix calculations , 1986 .

[11]  Davide Anguita,et al.  Fast training of Support Vector Machines for regression , 2000, Proceedings of the IEEE-INNS-ENNS International Joint Conference on Neural Networks. IJCNN 2000. Neural Computing: New Challenges and Perspectives for the New Millennium.

[12]  Jitendra Malik,et al.  Discriminative Decorrelation for Clustering and Classification , 2012, ECCV.

[13]  T. Chan An Optimal Circulant Preconditioner for Toeplitz Systems , 1988 .

[14]  Bill Triggs,et al.  Histograms of oriented gradients for human detection , 2005, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05).

[15]  W. Gragg,et al.  Superfast solution of real positive definite toeplitz systems , 1988 .

[16]  H. Akaike Block Toeplitz Matrix Inversion , 1973 .

[17]  Simon Lucey,et al.  Multi-channel Correlation Filters , 2013, 2013 IEEE International Conference on Computer Vision.

[18]  David Y. Y. Yun,et al.  Fast Solution of Toeplitz Systems of Equations and Computation of Padé Approximants , 1980, J. Algorithms.

[19]  B. V. Vijaya Kumar,et al.  Unconstrained correlation filters. , 1994, Applied optics.

[20]  Keinosuke Fukunaga,et al.  Introduction to statistical pattern recognition (2nd ed.) , 1990 .

[21]  Mário A. T. Figueiredo,et al.  Deconvolving Images With Unknown Boundaries Using the Alternating Direction Method of Multipliers , 2012, IEEE Transactions on Image Processing.

[22]  Raymond H. Chan,et al.  Conjugate Gradient Methods for Toeplitz Systems , 1996, SIAM Rev..

[23]  Alexei A. Efros,et al.  Unsupervised Discovery of Mid-Level Discriminative Patches , 2012, ECCV.

[24]  Peter H. N. de With,et al.  Fast Training of Object Detection Using Stochastic Gradient Descent , 2010, 2010 20th International Conference on Pattern Recognition.

[25]  Jonathon Shlens,et al.  Fast, Accurate Detection of 100,000 Object Classes on a Single Machine , 2013, 2013 IEEE Conference on Computer Vision and Pattern Recognition.

[26]  Takeo Kanade,et al.  Correlation Filters for Object Alignment , 2013, 2013 IEEE Conference on Computer Vision and Pattern Recognition.

[27]  Andrew E. Yagle,et al.  A fast algorithm for Toeplitz-block-Toeplitz linear systems , 2001, 2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.01CH37221).

[28]  Alexei A. Efros,et al.  Ensemble of exemplar-SVMs for object detection and beyond , 2011, 2011 International Conference on Computer Vision.

[29]  T. Kailath,et al.  Efficient inversion of Toeplitz-block Toeplitz matrix , 1983 .

[30]  David A. McAllester,et al.  Object Detection with Discriminatively Trained Part Based Models , 2010, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[31]  Stefano Serra Capizzano,et al.  Any Circulant-Like Preconditioner for Multilevel Matrices Is Not Superlinear , 2000, SIAM J. Matrix Anal. Appl..

[32]  Anders P. Eriksson,et al.  Fast Convolutional Sparse Coding , 2013, 2013 IEEE Conference on Computer Vision and Pattern Recognition.

[33]  Jiri Matas,et al.  Online learning of robust object detectors during unstable tracking , 2009, 2009 IEEE 12th International Conference on Computer Vision Workshops, ICCV Workshops.

[34]  François Fleuret,et al.  Accelerated Training of Linear Object Detectors , 2013, 2013 IEEE Conference on Computer Vision and Pattern Recognition Workshops.