Sabine Sternig

Sabine Sternig

Short CV

Sabine Sternig received her MSc Degree in Software Development and Business Management from the Graz University of Technology in December 2008. She is currently a research assistant at the Institute for Computer Graphics and Vision where she is currently involved in the MASA Project.


May 2011 Recipient of the 2011 Google Anita Borg Scholarship
February 2011 Workshop Chair of the 16th Computer Vision Winterworkshop held in Mitterberg, Austria
August 2010 Best Scientific Paper Award at the 20th International Conference on Pattern Recognition (ICPR) 2010, Istanbul, Turkey
Paper: Inverse Multiple Instance Learning for Classifier Grids
July 2009 Selected for oral presentation at International Computer Vision Summer School (ICVSS), Sicily


Current project
The aim of the project MASA is movement and action sequence analysis in complex sport games. In particular we are interested in team handball. The main goal is the identification of team tactics, discrimination of successful from non-successful player/team behavior, anticipation of player/team behavior and determination of physical demands during the game.

We are working on the multi-camera tracking task.

For more information see Projects

Finished project
The aim of the project Highway Monitoring is to investigate detection and tracking methods based on video and audio sensors, as well as the combination of these two modalities. To control the increasing traffic-flow on highways, monitoring of traffic is becoming more and more important. Existing incident detection systems shall be improved for the surveillance of highways. Starting from the existing methods and systems, the main goal of the proposed project is to improve and to adapt these systems in order to get more robust systems.

We worked on the vehicle detection task.

For more information see Projects


Inverse Multiple Instance Learning for Classifier Grids (Sternig, Roth, Bischof)

Recently, classifier grids have shown to be a considerable alternative for object detection from static cameras. However, one drawback of such approaches is drifting if an object is not moving over a long period of time. Thus, the goal of this work is to increase the recall of such classifiers while preserving their accuracy and speed. In particular, this is realized by adapting ideas from Multiple Instance Learning within a boosting framework. Since the set of positive samples is well defined, we apply this concept to the negative samples extracted from the scene: Inverse Multiple Instance Learning. By introducing temporal bags, we can ensure that each bag contains at least one sample having a negative label, providing the required stability. The experimental results demonstrate that using the proposed approach state-of-the-art detection results can by obtained, however, showing superior classification results in presence of non-moving objects.

TransientBoost (Sternig, Godec, Roth, Bischof)

For on-line learning algorithms, which are applied in many vision tasks such as detection or tracking, robust integration of unlabeled samples is a crucial point. Various strategies such as self-training, semi-supervised learning and multiple-instance learning have been proposed. However, these methods are either too adaptive, which causes drifting, or biased by a prior, which hinders incorporation of new (orthogonal) information. Therefore, we propose a new boosting-based on-line learning algorithm (TransientBoost), which is highly adaptive but still robust. This is realized by using an internal multi-class representation and modeling reliable and unreliable data in separate classes. Unreliable data is considered transient, hence we use highly adaptive learning parameters to adapt to fast changes in the scene while errors fade out fast. In contrast, the reliable data is preserved completely and not harmed by wrong updates. We demonstrate our algorithm on two different representative tasks, \ie, object detection and object tracking showing that we can handle typical problems considerable better than existing approaches. In addition, to demonstrate the stability and the robustness, we show long-term experiments for both tasks.
Longterm Tracking Results

Robust Adaptive Real-time Object Detection based on Classifier Grids

We introduce classifier grids in order to develop an adaptive but still robust real-time object detector for static cameras. Instead of using a sliding window for object detection we propose to train a separate classifier for each image location, obtaining a very specific object detector with a low false alarm rate. For each classifier corresponding to a grid element we estimate two generative representations in parallel, one describing the object's class and one describing the background. These are combined in order to obtain a discriminative model. To enable to adapt to changing environments these classifiers are learned on-line (i.e., boosting). Continuously learning (24 hours a day, 7 days a week) requires a stable system. In our method this is ensured by a fixed object representation while updating only the representation of the background. We demonstrate the stability in a long-term experiment by running the system for a whole week, which shows a stable performance over time. In addition, we compare the proposed approach to state-of-the-art methods in the field of person and car detection. In both cases we obtain competitive results.

ICVSS 2009 Poster
Download the ICVSS 2009 Poster


Longterm-dataset used in "Classifier Grids for Robust Adaptive Object Detection", CVPR'09: Download


Download Results Caviar Sequence

Download Results Pets Sequence

Download Results Highway Sequence

Download Results Corridor Sequence



  1. Hough Forests Revisited: An Approach to Multiple Instance Tracking from Multiple Cameras (bib)Georg Poier, Samuel Schulter, Sabine Sternig, Peter M. Roth, and Horst Bischof In Proc. German Conference on Pattern Recognition (GCPR/DAGM), 2014
    (The original publication is available at


  1. Robust Real-Time Tracking of Multiple Objects by Volumetric Mass Densities (bib) (data) Horst Possegger, Sabine Sternig, Thomas Mauthner, Peter M. Roth, and Horst Bischof In Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2013


  1. Unsupervised Calibration of Camera Networks and Virtual PTZ Cameras (bib) (code) (data) Horst Possegger, Matthias RĂ¼ther, Sabine Sternig, Thomas Mauthner, Manfred Klopschitz, Peter M. Roth, and Horst Bischof In Proc. Computer Vision Winter Workshop (CVWW), 2012
  2. Hough Regions for Joining Instance Localization and Segmentation (bib)Hayko Riemenschneider, Sabine Sternig, Michael Donoser, Peter M. Roth, and Horst Bischof In Proc. European Conference on Computer Vision (ECCV), 2012
  3. On-line Inverse Multiple Instance Boosting for Classifier Grids (bib)Sabine Sternig, Peter M. Roth, and Horst Bischof Pattern Recognition Letters, 2012


  1. Multi-camera Multi-object Tracking by Robust Hough-based Homography Projections (bib)Sabine Sternig, Thomas Mauthner, Arnold Irschara, Peter M. Roth, and Horst Bischof In Proc. 11th IEEE Workshop on Visual Surveillance (ICCV), 2011
  2. Proc. 16th Computer Vision Winter Workshop (bib)Andreas Wendel, Sabine Sternig, and Martin Godec, eds. Verlag der Technische Universitaet Graz, 2011


  1. Context-driven Clustering by Multi-class Classification in an Active Learning Framework (bib)Martin Godec, Sabine Sternig, Peter M. Roth, and Horst Bischof In Proc. Workshop on Use of Context in Video Processing (CVPR), 2010
  2. TransientBoost: On-line Boosting with Transient Data (bib)Sabine Sternig, Martin Godec, Peter M. Roth, and Horst Bischof In Proc. IEEE Online Learning for Computer Vision Workshop (CVPR), 2010
  3. Robust object detection by classifier cubes and local verification (bib)Sabine Sternig, Hayko Riemenschneider, Peter M. Roth, Michael Donoser, and Horst Bischof In Proc. Workshop of the Austrian Association for Pattern Recognition, 2010
  4. Inverse Multiple Instance Learning for Classifier Grids (bib)Sabine Sternig, Peter M. Roth, and Horst Bischof In Proc. IEEE Int'l Conf. on Pattern Recognition, 2010
    (Winner of the Best Paper Award)
  5. Learning of Scene-Specific Object Detectors by Classifier Co-Grids (bib)Sabine Sternig, Peter M. Roth, and Horst Bischof In Proc. IEEE Int'l Conf. on Advanced Video and Signal-Based Surveillance, 2010


  1. Classifier Grids for Robust Adaptive Object Detection (bib)Peter M. Roth, Sabine Sternig, Helmut Grabner, and Horst Bischof In Proc. IEEE Conf. on Computer Vision and Pattern Recognition, 2009
  2. Robust Adaptive Classifier Grids for Object Detection from Static Cameras (bib)Sabine Sternig, Peter M. Roth, Helmut Grabner, and Horst Bischof In Proc. Computer Vision Winter Workshop, 2009

Copyright 2010 ICG

Valid XHTML 1.1 Valid CSS