Till Quack
work in computer vision

currently

I am a post-graduate researcher at the Computer Vision Lab at ETH Zurich. My research interests include object recognition, multimedia mining, image retrieval, applications of computer vision on mobile devices and the web.

recently

publicly

Publications in Conference Proceedings and Journals

1. Stephan Gammeter, Lukas Bossard, Till Quack and Luc Van Gool
   I know what you did last summer: object-level auto-annotation of holiday snaps (to appear)
   ICCV 2009, Kyoto, Japan, 27.9.2009 - 4.10.2009
2. Luc Van Gool, Michael Breitenstein, Stephan Gammeter, Helmut Grabner, Till Quack
   Mining from large image sets (Keynote address)
   CIVR 2009, Santorini, Greece, 8.7.2009 - 11.7.2009
3. Till Quack
   Large-scale Mining and Retrieval of Visual Data in a Multimodal Context (pdf download soon)
   PHD Thesis No 18910, ETH Zurich, Jan 2009.
4. Till Quack, Bastian Leibe and Luc Van Gool
   World-scale Mining of Objects and Events from Community Photo Collections
   CIVR 2008, Niagara Falls, Canada, 7.-9. July 2008
5. Till Quack, Herbert Bay and Luc Van Gool
   Object Recognition for the Internet of Things
   Internet of Things 2008, Zurich, Switzerland
6. Till Quack, Vittorio Ferrari, Bastian Leibe and Luc Van Gool
   Efficient Mining of Frequent and Distinctive Feature Configurations
   ICCV 2007, Rio de Janeiro, Brazil
7. Till Quack, Vittorio Ferrari, Luc Van Gool,
   Video Mining with Frequent itemset Configurations.
   CIVR 2006, Tempe, AZ, USA, July 2006
8. Till Quack, Ullrich Mönich, Lars Thiele, B.S. Manjunath,
   "Cortina: A System for Large-scale, Content-based Web Image Retrieval"
   ACM Multimedia 2004, New York, NY, USA, Oct. 2004.

Invited Talks

1. Till Quack, Cross-Modal Mining of Objects and Events from Community Photo Collections, 13.7.2009, IJCAI-09 Workshop on Cross-Media Information Access and Mining, Pasadena, CA, USA
2. Till Quack, Scalable Mining of Objects and Events from Community Photo Collections, 6.11.2008, Autumn 2008 Pattern Recognition and Computer Vision Colloquium, The Center for Machine Perception at Czech Technical University, Prague, Czech Republic
3. Till Quack, Frequent Itemset Mining and Association Rules in Visual Data, 23.10.2007, UCLA IPAM Mathematics of Knowledge and Search Engines Workshops II: Numerical Tools and Fast Algorithms for Massive Data Mining, Search Engines and Applications, Los Angeles, CA, USA
4. Till Quack, Local Features for Mining Multimedia Data and Connecting the Real and Digital World, 26.10.2007, UCSB Interactive Digital Media, NSF IGERT Seminar Series, Santa Barbara, CA, USA

Patents

1. "Method and System for Image-based Information Retrieval", 2007.
2. "Method and system for generating a pictorial reference database", 2008.

supervisedly

I defined and supervised the following student projects:

1. Augmented Reality on the iPhones (working title). Aleksander Slater. WS 08/09. MSc project.
2. Optimized features for Logo Recognition (working title). Wouter Vandenbussche. WS 08/09. MSc project.
3. Optimized visual vocabularies for large-scale auto-annotation of landmark buildings. Lukas Bossard. WS 08/09. MSc project.
4. Unsupervised 3D reconstruction from images mined in community photo collections. Fabio Magagna. SS 08. MSc project.
5. A novel, motion-detection based user interface for mobile visual search. Marcel Breu and Marc Mueller. SS 08. Semester Project.
6. Detecting and Reading Text in Natural Scenes. Martin Renold, WS/SS 07/08. MSc Project
7. Real-time Server-side Object Recognition from Mobile Devices. Raphael Jecker and Benjamin Knecht, WS 2007. Semester Project
8. Using Mobile Object Recognition to Tag Slides in Meeting Rooms. Dominik Angehrn, WS 2007. Semester Project
9. High Performance Object Recognition. David Scheiner, Reto Schwarz. SS 2007. MSc project.
10. Graph Mining for Object Class Recognition. Sarah Gugl. SS 2007. MSc project.
11. A Cityguide on a mobile phone with Object Recognition and Localization. WS 2006/2007. Fabio Magagna. BSc project.
12. Implementation of a Vocabulary Tree for fast Feature Matching. WS 2006/2007. Remo Poretti. Semester project.
13. SURF on a Mobile Phone. Tamara Ullrich. SS 2006. Semester project.
14. SIFT on a Mobile Phone. Marcel Germann.WS 2005/2006. Semester project.
15. Abstracting Music Video Clips automatically. Philippe Wüger. SS 2005. Semester project. (co-supervised with Pascal Mueller)

commercially

In 2006 I co-founded kooaba. Kooaba offers mobile visual search (or object recognition, for that matter) and is a direct consequence from research here at the Computer Vision Lab at ETH.

In 1998 I founded Quack Internet Solutions. Quack Internet Solutions offers web-site design, programming and related services like consulting or on-line marketing. In recent years I put a strong focus on standards-based web-development with CSS.

previously

Before that I was involved in the project Cortina, a system for large-scale, content-based image retrieval on the World Wide Web. It came into existence during a stay in the Vision Lab at the University of California at Santa Barbara, USA. My Master's thesis discusses the system and looks into the semantics within.

Master's Thesis: Cortina (2004)

You can check out the system here: www.cortina.ch. (Note: Since it is still a research system, it might be offline every now and then).

Download the Master's thesis as a PDF file: Cortina, a System for Large-scale, Content-based Web Image-retrieval - and the Semantics within. (1.1MB, 101 pages)

Markerless 3D Augmented Reality (2003)

Lukas Hohl and yours truly implemented a system for 3D-augmentation of video sequences. 3D information was obtained from non-coplanar regions, tracked with Vittorio Ferraris real-time affine region tracker. This was a semester-project in the Computer Vision Lab of ETH Zurich back in 2003.

• Tram scene (2d Augmentation) (The number 4 is mapped to the tram)
• Buddha Scene 1 (3d Augmentation)
• Buddha Scene 2 (3d Augmentation)
• The tracked regions for the Buddha scene
  (hard to see, look at the fine green lines at the edge of the book pile)
• The 3D Bounding Box for the Buddha scene
  (hard to see, look at the fine white grid-box on top of the book pile)
• A coke can in the office scene (3D Augmentation)
• A 2D Augmentation with heavy illumination changes

• Report in PDF Format (628kB, 29 pages)

Combination of Space-Time Codes with Forward Error Correction (FEC)

Not exactly computer vision, but for reasons of completeness: The other semster-project, created in cooperation with Tobias "Tobi" Koch.

• Report in PDF Format (1.24 MB, 74 pages)

personally

Till Quack
Computer Vision Laboratory
Sternwartstrasse 7
ETH-Zentrum
CH - 8092 Zürich, Switzerland

Phone: +41 1 63 27632
Fax: +41 1 63 21199
Office: ETF D114.1
E-Mail: tquack AT ee.ethz.ch

weblog

Quack IA weblog