Navigation

Martin Köstinger

Martin Köstinger

Short CV

Martin Köstinger received his MSc Degree (Dipl.-Ing.) in Software Development and Business Management and a PhD degree (Dr.techn) in Computer Science, both from the Graz University of Technology in 2009 and 2013, respectively. He is currently a research assistant at the Institute for Computer Graphics and Vision where he is involved in the FACTS Project. His research is focused on Object Detection and Recognition, with a special focus on faces.

News:

Research

Large Scale Metric Learning from Equivalence Constraints, CVPR 2012

(Koestinger, Hirzer, Wohlhart, Roth, Bischof)

In this work, we raise important issues on scalability and the required degree of supervision of existing Mahalanobis metric learning methods. Often rather tedious optimization procedures are applied that become computationally intractable on a large scale. Further, if one considers the constantly growing amount of data it is often infeasible to specify fully supervised labels for all data points. Instead, it is easier to specify labels in form of equivalence constraints. We introduce a simple though effective strategy to learn a distance metric from equivalence constraints, based on a statistical inference perspective. In contrast to existing methods we do not rely on complex optimization problems requiring computationally expensive iterations. Hence, our method is orders of magnitudes faster than comparable methods. Results on a variety of challenging benchmarks with rather diverse nature demonstrate the power of our method. These include faces in unconstrained environments, matching before unseen object instances and person re-identification across spatially disjoint cameras. In the latter two benchmarks we clearly outperform the state-ofthe- art.

The work was supported by the Austrian Science Foundation (FWF) project Advanced Learning for Tracking and Detection in Medical Workflow Analysis (I535-N23) and by the Austrian Research Promotion Agency (FFG) project SHARE in the IV2Splus program.

Full text (pdf), POSTER (png), Code (Matlab)

Synergy-based Learning of Facial Identity, DAGM-OEAGM 2012

(Koestinger, Roth, Bischof)

In this work we address the problem that most face recognition approaches neglect that faces share strong visual similarities, which can be exploited when learning discriminative models. Hence, we propose to model face recognition as multi-task learning problem. This enables us to exploit both, shared common information and also individual characteristics of faces. In particular, we build on Mahalanobis metric learning, which has recently shown good performance for many computer vision problems. Our main contribution is twofold. First, we extend a recent efficient metric learning algorithm to multi-task learning. The resulting algorithm supports label-incompatible learning which allows us to tap the rather large pool of anonymously labeled face pairs also for face identification. Second, we show how to learn and combine person specific metrics for face identification improving the classification power. We demonstrate the method for different face recognition tasks where we are able to match or slightly outperform state-of-the-art multi-task learning approaches.

The work was supported by the Austrian Science Foundation (FWF) project Advanced Learning for Tracking and Detection in Medical Workflow Analysis (I535-N23) and by the Austrian Research Promotion Agency (FFG) project SHARE in the IV2Splus program.

Full text (pdf)

Robust Face Detection by Simple Means, CVAW (DAGM-OEAGM) 2012

(Koestinger, Wohlhart, Roth, Bischof)

Face detection is still one of the core problems in computer vision, especially in unconstrained real-world situations where variations in face pose or bad imaging conditions have to be handled. These problems are covered by recent benchmarks such as Face Detection Dataset and Benchmark (FDDB) [Jain and Learned-Miller, 2010], which reveals that established methods, e.g, Viola and Jones [Viola and Jones, 2001] suffer a drop in performance. More effctive approaches exist, but are closed source and not publicly available. Thus, we propose a simple but effective detector that is available to the public. It combines Histograms of Orientated Gradient (HOG) [Dalal and Triggs, 2005] features with linear Support Vector Machine (SVM) classiffication.

The work was supported by the Austrian Science Foundation (FWF) project Advanced Learning for Tracking and Detection in Medical Workflow Analysis (I535-N23) and by the Austrian Research Promotion Agency (FFG) project SHARE in the IV2Splus program.

Full text (pdf), Code available soon (C++)

Videos

Face Recognition

... from Videos only using Weakly Related Information Cues

Visual Speaker Identification

Face Detection

Optical Character Recognition in Videos (VideoOCR) and Natural Images

Source: OCR Image with inpainted text:OCR

Student Projects


TBA

Publications

2014

  1. Mahalanobis Distance Learning for Person Re-Identification (bib)Peter M. Roth, Martin Hirzer, Martin Koestinger, Csaba Beleznai, and Horst Bischof In Person Re-Identification, pages 247-267, Springer, 2014
    (The original publication is available at www.springer.com)

2013

  1. Joint Learning of Discriminative Prototypes and Large Margin Nearest Neighbor Classifiers (bib)Martin Koestinger, Paul Wohlhart, Peter M. Roth, and Horst Bischof In Proc. International Conference on Computer Vision (ICCV), 2013
  2. Efficient Retrieval for Large Scale Metric Learning (bib)Martin Koestinger, Peter M. Roth, and Horst Bischof In Proc. German Conference on Pattern Recognition (GCPR/DAGM), 2013
  3. Optimizing 1-Nearest Prototype Classifiers (bib)Paul Wohlhart, Martin Koestinger, Michael Donoser, Peter M. Roth, and Horst Bischof In Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2013

2012

  1. Dense Appearance Modeling and Efficient Learning of Camera Transitions for Person Re-Identification (bib)Martin Hirzer, Csaba Beleznai, Martin Koestinger, Peter M. Roth, and Horst Bischof In Proc. IEEE International Conference on Image Processing (ICIP), 2012
  2. Relaxed Pairwise Learned Metric for Person Re-Identification (bib)Martin Hirzer, Peter M. Roth, Martin Koestinger, and Horst Bischof In Proc. European Conference on Computer Vision (ECCV), 2012
    (The original publication is available at www.springerlink.com)
  3. Large Scale Metric Learning from Equivalence Constraints (bib)Martin Koestinger, Martin Hirzer, Paul Wohlhart, Peter M. Roth, and Horst Bischof In Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2012
  4. Synergy-based Learning of Facial Identity (bib)Martin Koestinger, Peter M. Roth, and Horst Bischof In Proc. DAGM Symposium, 2012
    (Winner of the Best Paper Award)
  5. Robust Face Detection by Simple Means (bib)Martin Koestinger, Paul Wohlhart, Peter M. Roth, and Horst Bischof In Computer Vision in Applications Workshop (DAGM), 2012
  6. Discriminative Hough Forests for Object Detection (bib)Paul Wohlhart, Samuel Schulter, Martin Koestinger, Peter M. Roth, and Horst Bischof In Proc. British Machine Vision Conference (BMVC), 2012

2011

  1. Open Source Intelligence am Beispiel von KIRAS MDL: Multimedia Documentation Lab (bib)Gerhard Backfried, Dorothea Aniola, Gerald Quirchmayr, Werner Winiwarter, Klaus Mak, H. C. Pilles, Christian Meurers, Martin Koestinger, Paul Wohlhart, and Peter M. Roth In Proceedings of 9. Sicherheitskonferenz Krems, Donau-Universitaet Krems, 2011
  2. Learning to Recognize Faces from Videos and Weakly Related Information Cues (bib)Martin Koestinger, Paul Wohlhart, Peter M. Roth, and Horst Bischof In Proc. IEEE International Conference on Advanced Video and Signal-Based Surveillance (AVSS), 2011
  3. Annotated Facial Landmarks in the Wild: A Large-scale, Real-world Database for Facial Landmark Localization (bib)Martin Koestinger, Paul Wohlhart, Peter M. Roth, and Horst Bischof In First IEEE International Workshop on Benchmarking Facial Image Analysis Technologies, 2011
  4. Multiple Instance Boosting for Face Recognition in Videos (bib)Paul Wohlhart, Martin Koestinger, Peter M. Roth, and Horst Bischof In Proc. DAGM Symposium, 2011
  5. Learning Face Recognition in Videos from Associated Information Sources (bib)Paul Wohlhart, Martin Koestinger, Peter M. Roth, and Horst Bischof In Proc. Workshop of the Austrian Association for Pattern Recognition (AAPR/OAGM), 2011

2010

  1. Video Detection of Dangerous Goods Vehicles in Road Tunnels (bib)Josef Birchbauer, Martin Koestinger, Paul Wohlhart, Peter M. Roth, Horst Bischof, and Claudia Windisch In Proc. Tunnel Safety and Ventilation, 2010
  2. Automatic Detection and Reading of Dangerous Goods Plates (bib)Peter M. Roth, Martin Koestinger, Paul Wohlhart, Horst Bischof, and Josef Birchbauer In Proc. IEEE Int'l Conf. on Advanced Video and Signal-Based Surveillance, 2010

2010

  1. Planar Trademark and Logo Retrieval (bib)Martin Koestinger, Peter M. Roth, and Horst Bischof Technical Report, Graz University of Technology, Inst. f. Computer Graphics and Vision, ICG-TR-10/01, 2010 (Presented at Computer Vision Winter Workshop 2010)

2009

  1. KIRAS-MDL State-of-the-Art Report (bib)Martin Koestinger, Paul Wohlhart, Peter M. Roth, and Horst Bischof Technical Report, Graz University of Technology, Inst. f. Computer Graphics and Vision, ICG-TR-09/09, 2009

2013

  1. Efficient Metric Learning for Real-World Face Recognition (bib)Martin Koestinger Ph.D. Thesis, Graz University of Technology, Faculty of Computer Science, 2013

2009

  1. An Object Recognition System for planar Trademark and Logo Retrieval (bib)Martin Koestinger MSc. Thesis, Graz University of Technology, Faculty of Computer Science, 2009






Copyright 2010 ICG

Valid XHTML 1.1 Valid CSS