Abstract
The problem of learning the class identity of visual objects has received considerable attention recently. With rare exception, all of the work to date assumes low variation in appearance, which limits them to a single depictive style usually photographic. The same object depicted in other styles — as a drawing, perhaps — cannot be identified reliably. Yet humans are able to name the object no matter how it is depicted, and even recognise a real object having previously seen only a drawing.
This paper describes a classifier which is unique in being able to learn class identity no matter how the class instances are depicted. The key to this is our proposition that topological structure is a class invariant. Practically, we depend on spectral graph analysis of a hierarchical description of an image to construct a feature vector of fixed dimension. Hence structure is transformed to a feature vector, which can be classified using standard methods. We demonstrate the classifier on several diverse classes.
Chapter PDF
Similar content being viewed by others
Keywords
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.
References
Ahuja, N., Todorovic, S.: Discovering hierarchical taxonomy of categories and shared subcategories in images. In: ICCV (2007)
Leibe, B., Schiele, B.: Interleaved object categorization and segmentation. In: BMVC
Andrew Bangham, J., Moravoc, K., Harvey, R., Fisher, M.: Scale-space trees and applications as filters, for stereo vision and image retrieval. In: Pridmore, T., Elliman, D. (eds.) British Machine Vision Conference (August 1999)
Chung, F.R.K.: Spectral graph theory. American Mathematical Society (1997)
Lowe, D.: Distinctive image features from scale-invariant keypoints. International Journal of Computer Vision 1, 91–110 (2004)
Fidler, S., Leonardis, A.: Towards scalable representations of object categories: Learning a hierarchy of parts. In: ICCV (2007)
Schroff, F., Criminisi, A., Zisserman, A.: Havesting image databases from the web. In: ICCV (2007)
Grauman, K., Darrell, T.: Unsupervised learning of catagories from sets of partially matching image features. In: CVPR (2006)
Gutman, I., Zhou, B.: Laplacian energy of a graph. Linear Algebra and its Applications 44, 29–37 (2006)
Mikolajczyk, K., Schmid, C.: A performance evaluation of local descriptors. TPAMI 2005 27(10), 1615–1630 (2005)
Fei-Fei, L., Fergus, R., Perona, P.: A bayesian approach to unsupervised one-shot learning. In: ICCV (2003)
Matas, J., Chum, O., Urban, M., Pajdla, T.: Robust wide baseline stereo from maximally stable extremal regions. In: BMVC (2002)
Mikolajczyk, K., Leibe, B., Schiele, B.: Multiple object class detection with a generative model. In: CVPR (2006)
Mutch, J., Lowe, D.: Multiclass object recognition with sparse, localized features. In: CVPR (2006)
Pinz, A., Opelt, A., Zisserman, A.: A boundary-fragment-model for object detection. In: ECCV (2006)
Yan, P., Khan, S.M., Shah, M.: 3d model based object detection in an arbitrary view. In: ICCV (2007)
Shi, J., Malik, J.: Normalized cuts and image segmentation. In: International Conference on Computer Vision and Pattern Recognition (1997)
Schechtman, S., Irani, M.: Matching local self-similarities across images and videos. In: CVPR (2007)
Weber, M., Welling, M., Perona, P.: Unsupervised learning of models for visual object class recognition
Zhu, P., Wilson, R.C.: Stability of the eigenvalues of graphs. In: Gagalowicz, A., Philips, W. (eds.) CAIP 2005. LNCS, vol. 3691, pp. 371–378. Springer, Heidelberg (2005)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2008 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Xiao, B., Song, YZ., Balika, A., Hall, P.M. (2008). Structure Is a Visual Class Invariant. In: da Vitoria Lobo, N., et al. Structural, Syntactic, and Statistical Pattern Recognition. SSPR /SPR 2008. Lecture Notes in Computer Science, vol 5342. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-540-89689-0_37
Download citation
DOI: https://doi.org/10.1007/978-3-540-89689-0_37
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-89688-3
Online ISBN: 978-3-540-89689-0
eBook Packages: Computer ScienceComputer Science (R0)