SPECIAL ISSUE: New Trends in Artificial Vision and Artificial Intelligence; Edited by M. De Gregorio and M. FrucciNo Access

NEURAL OBJECT RECOGNITION BY HIERARCHICAL APPEARANCE LEARNING

DANIEL OBERHOFF

FIT - LIFE, Fraunhofer Society, Schloss Birlinghoven, 53754 St. Augustin, Germany

Search for more papers by this author

and

MARINA KOLESNIK

FIT - LIFE, Fraunhofer Society, Schloss Birlinghoven, 53754 St. Augustin, Germany

Search for more papers by this author

https://doi.org/10.1142/S0218001408006582Cited by:0 (Source: Crossref)

Abstract

We present a system for object recognition that is largely inspired by physiologically identified processing streams in the visual cortex, specifically in the ventral stream. It consists of neural units organized in a hierarchy of layers with encoding features of increasing complexity. A key feature of the system is that the neural units learn their preferred patterns from visual input alone. Through this "soft wiring" of neural units the system becomes tuned for target object classes through passive visual experience and no labels are required in this stage. Object labels are only introduced in the last step to train a classifier on the system's output. While this tuning process is purely feed-forward we also present a neural mechanism for back projection of the learned image patterns down the hierarchical layers, and demonstrate how this feedback can be used to stabilize the system in the presence of noise. We test the neural system with natural images from publicly available data-sets of natural scenes and handwritten digits.

This work is partially supported under the MCCOOP EU NEST-2003-1 Adventure project Contract no.: 012963-2.

Keywords:

References

G. Deco and E. T. Rolls, Vis. Res. 44, (2004). Google Scholar
K. Fukushima, Biol. Cybern. 36(4), 193 (1980), DOI: 10.1007/BF00344251. Crossref, Web of Science, Google Scholar
S. Grossberg and R. D. S. Raizada, Vis. Res. 40(10), 1413 (1999), DOI: 10.1016/S0042-6989(99)00229-1. Crossref, Web of Science, Google Scholar
R. Hadsell, S. Chopra and Y. LeCun, Dimensionality reduction by learning an invariant mapping, Proc. CVPR (IEEE Press, 2006) pp. 1735–1742. Google Scholar
D. H. Hubel and T. N. Wiesel, J. Physiol. 195, (1967). Google Scholar
C. P. Hunget al., Sci. Rep. 863 (2005). Google Scholar
E. Kobatake and K. Tanaka, J. Neurophys. 71, 856 (1994). Crossref, Web of Science, Google Scholar
Y. LeCunet al., IEEE Commun. 41 (1989), DOI: 10.1109/35.41400. Google Scholar
S. Munder and D. M. Gavrila, IEEE Trans. Patt. Anal. Mach. Intell. 28, 1863 (2006), DOI: 10.1109/TPAMI.2006.217. Crossref, Web of Science, Google Scholar
J. Mutch and D. G. Lowe, Multiclass object recognition with sparse, localized features, Proc. CVPR (2006) pp. 11–18. Google Scholar
D. Oberhoff and M. Kolesnik, Adv. Brain Vis. Artif. Intell. (2007). Google Scholar
M. Riesenhuber and T. Poggio, Computational Object Vision 1640 (2003). Google Scholar
M. Riesenhuber and T. Poggio, Nat. Rev. (2000). Google Scholar
E. T. Rolls and G. Deco , The Computational Neuroscience of Vision ( Oxford University Press , 2002 ) . Google Scholar
T. Serre, M. Kouh, C. Cadieu, U. Knoblich, T. Poggio and G. Kreiman, A theory of object recognition: computations and circuits in the feedforward path of the ventral stream in primate visual cortex, CBCL Memo 259 (2005) . Google Scholar
A. Tschieler and H. Neumann, Neuroscience 122, 921 (2003). Web of Science, Google Scholar
J. K. Tsotsoset al., A. I. 78, 507 (1994). Google Scholar
L. G. Ungerleider and M. Mishkin, Analysis of Visual Behavior (MIT Press, 1992) pp. 549–586. Google Scholar
H. Zhanget al., Svm-knn: discriminative nearest neighbor classication for visual category recognition, Proc. CVPR2 (2006) pp. 2126–2136. Google Scholar

Vol. 22, No. 05

Metrics

Downloaded 22 times

History

Keywords

PDF download

System Upgrade on Tue, May 28th, 2024 at 2am (EDT)

NEURAL OBJECT RECOGNITION BY HIERARCHICAL APPEARANCE LEARNING

Abstract

Recommended