Self-Adaptable Templates for Feature Coding

Hierarchical feed-forward networks have been successfully applied in object recognition. At each level of the hierarchy, features are extracted and encoded, followed by a pooling step. Within this processing pipeline, the common trend is to learn the feature coding templates, often referred as codebook entries, filters, or over-complete basis. Recently, an approach that apparently does not use templates has been shown to obtain very promising results. This is the second-order pooling (O2P) [1, 2, 3, 4, 5]. In this paper, we analyze O2P as a coding-pooling scheme. We find that at testing phase, O2P automatically adapts the feature coding templates to the input features, rather than using templates learned during the training phase. From this finding, we are able to bring common concepts of coding-pooling schemes to O2P, such as feature quantization. This allows for significant accuracy improvements of O2P in standard benchmarks of image classification, namely Caltech101 and VOC07.

[1]  Chih-Jen Lin,et al.  LIBLINEAR: A Library for Large Linear Classification , 2008, J. Mach. Learn. Res..

[2]  Luc Van Gool,et al.  Sparse Quantization for Patch Description , 2013, 2013 IEEE Conference on Computer Vision and Pattern Recognition.

[3]  J. Weickert,et al.  Visualization and Processing of Tensor Fields (Mathematics and Visualization) , 2005 .

[4]  T. Poggio,et al.  Hierarchical models of object recognition in cortex , 1999, Nature Neuroscience.

[5]  Thomas Mensink,et al.  Image Classification with the Fisher Vector: Theory and Practice , 2013, International Journal of Computer Vision.

[6]  Luc Van Gool,et al.  The Pascal Visual Object Classes (VOC) Challenge , 2010, International Journal of Computer Vision.

[7]  Hans Hagen,et al.  Visualization and Processing of Tensor Fields , 2014 .

[8]  Kunihiko Fukushima,et al.  Neocognitron: A self-organizing neural network model for a mechanism of pattern recognition unaffected by shift in position , 1980, Biological Cybernetics.

[9]  Cristian Sminchisescu,et al.  Semantic Segmentation with Second-Order Pooling , 2012, ECCV.

[10]  Qilong Wang,et al.  Local Log-Euclidean Covariance Matrix (L2ECM) for Image Representation and Its Applications , 2012, ECCV.

[11]  Nicholas Ayache,et al.  Geometric Means in a Novel Vector Space Structure on Symmetric Positive-Definite Matrices , 2007, SIAM J. Matrix Anal. Appl..

[12]  Fatih Murat Porikli,et al.  Region Covariance: A Fast Descriptor for Detection and Classification , 2006, ECCV.

[13]  R. Bhatia.  Positive Definite Matrices , 2007 .

[14]  Gabriela Csurka,et al.  Visual categorization with bags of keypoints , 2002, eccv 2004.

[15]  Pietro Perona,et al.  One-shot learning of object categories , 2006, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[16]  David D. Cox,et al.  Making a Science of Model Search: Hyperparameter Optimization in Hundreds of Dimensions for Vision Architectures , 2013, ICML.

[17]  Thomas S. Huang,et al.  Image Classification Using Super-Vector Coding of Local Image Descriptors , 2010, ECCV.

[18]  Jean Ponce,et al.  A graph-matching kernel for object categorization , 2011, 2011 International Conference on Computer Vision.

[19]  G LoweDavid,et al.  Distinctive Image Features from Scale-Invariant Keypoints , 2004 .

[20]  Andrew Y. Ng,et al.  The Importance of Encoding Versus Training with Sparse Coding and Vector Quantization , 2011, ICML.

[21]  Trevor Darrell,et al.  Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation , 2013, 2014 IEEE Conference on Computer Vision and Pattern Recognition.

[22]  Luc Van Gool,et al.  Nested Sparse Quantization for Efficient Feature Coding , 2012, ECCV.

[23]  D. Le Bihan,et al.  Diffusion tensor imaging: Concepts and applications , 2001, Journal of magnetic resonance imaging : JMRI.