Machine Learning

Pattern Learning and Recognition in Vision

(Click animations to start playback)

The salient components of visual scenes are objects. While our visual system identifies visual objects with apparent ease, this task, in general, is still considered as one of the major unsolved problems in computer vision. In the development of component extraction algorithms specific for visual data, we address the problem of autonomous object learning. In my group and together with our collaborators we develop generative approaches incorporating explicit object occlusions and models that consider object invariances. The models are developed for challenging settings of cluttered scenes. Using probabilistic and neuro-dynamic approaches we develop applications to autonomous pattern learning and visual pattern recognition. To accomplish these tasks, our algorithms are combined with established procedures and can be combined with segmentation or methods for motion field extraction, and they are trained using grounded methods for the integration of discriminative and generative approaches.

Further Reading

  • Z. Dai and J. Lücke (2014).
    Autonomous Document Cleaning – A Generative Approach to Reconstruct Strongly Corrupted Scanned Texts.
    IEEE Transactions on Pattern Analysis and Machine Intelligence, 36(10): 1950-1962. (online accessbibtex)

  • Z. Dai and J. Lücke (2012)
    Autonomous Cleaning of Corrupted Scanned Documents - A Generative Modeling Approach (pdf, bibtex)
    Proc. IEEE Computer Vision and Pattern Recognition (CVPR), 3338-3345, oral presentation.
    (highest CVPR 2012 reviewer score and Google Student Travel Award)

  • Z. Dai and J. Lücke (2012)
    Unsupervised Learning of Translation Invariant Occlusive Components (pdf, bibtex)
    Proc. IEEE Computer Vision and Pattern Recognition (CVPR), 2400-2407.

  • C. Keck and J. Lücke (2010).
    Learning of Lateral Connections for Representational Invariant Recognition (pdf, bibtex).
    Proc. ICANN 2010, LNCS 6354, 21-30.

  • C. Keck, J. D. Bouecke, and J. Lücke (2009).
    J. D. Bouecke and J. Lücke (2008).
    Learning of Neural Information Routing for Correspondence Finding (pdf, bibtex).
    Proc. ICANN, Springer, LNCS 5164, 557-566.

  • P. Wolfrum, C. Wolff, J. Lücke, and C. von der Malsburg (2008).
    A Recurrent Dynamic Model for Correspondence-Based Face Recognition (pdf, bibtex).
    Journal of Vision 8(7):34, 1-18.

  • J. Lücke, C. Keck, and C. von der Malsburg (2008).
    Rapid Convergence to Feature Layer Correspondences (preprint, doi, bibtex).
    Neural Computation 20(10):2441-2463.

  • J. Lücke and C. von der Malsburg (2006).
    Rapid Correspondence Finding in Networks of Cortical Columns (pdf, bibtex).
    Proc. ICANN, Springer, LNCS 4131, 668-677.

     

Copyright notice

The papers listed above have been published after peer review in different journals. These journals remain the only definitive repository of the content. Copyright and all rights therein are usually retained by the respective publishers. These materials may not be copied or reposted without their explicit permission. Use for scholarly purposes only.