Perception in cognitive systems

There are no prerequisites.

Lectures:

• Computational theories of perception
• Cognitive architectures of perception
• Learning, recognition, categorization and abstractions of visual entities
• Active vision
• Computational theories of attentional mechanisms
• Visual context
• Computational theories of spatial perception
  Exercises:
  Practical implementation of computational models related to perception and cognition. Under supervision, development of software and hardware solutions for object recognition and categorisation, robot localisation, and active vision.

1. Object Categorization: Computer and Human Vision Perspectives, S. J. Dickinskon, A. Leonardis, B. Schiele, M. J. Tarr, (Eds.), Cambridge University Press, 2009, (ISBN-13: 9780521887380).
2. A. Pinz, Object Categorization, Foundations and Trends® in Computer Graphics and Vision, 1(4), pp. 255-353, 2006, (ISBN: 1-933019-13-1).
   Dostopna tudi: http://www.emt.tugraz.at/system/files/CGV003-journal.pdf
3. S. Thrun, W. Burgard, D. Fox, Probabilistic Robotics: Intelligent Robotics and Autonomous Agents, (ISBN-10: 0262201623).

The objective of the course is to teach the students basic competences in the area of artificial perception in cognitive systems, including selected computational theories of perception, computational models of perceptual processes, and application of these models for designing active cognitive robotic systems.

After successfully completing the course, the students will be able to:

-understand computational models of perception and their implementation in artificial cognitive systems,
- understand design principles for practical problems in the area of artificial perception in cognitive systems,
- design and implement practical solutions in the area of machine perception in cognitive systems, e.g., in autonomous robots, control systems, intelligent environments or mobile computing,
- understand wider research area of artificial and natural perception and cognitive systems,
- perform research based on professional literature and experimental work and program sensorial and robot systems.

Lectures with slides. Exercises in appropriately equipped laboratories. Individual work and work in small groups.

Continuing (homework, midterm exams, project work)
Final (written and oral exam)
grading: 5 (fail), 6-10 (pass) (according to the Statute of UL)

Pet najpomembnejših del:
1. A. Leonardis, A. Gupta, and R. Bajcsy, »Segmentation of range images as the search for geometric parametric models«, International Journal of Computer Vision, 14, pages 253-277, 1995.
2. A. Leonardis, A. Jaklic, and F. Solina, »Superquadrics for segmentation and modelling range data«, IEEE Transactions on Pattern Analysis and Machine Intelligence, 19, pages 1289-1295, 1997.
3. A. Leonardis and H. Bischof, »Robust recognition using eigenimages«, Computer Vision and Image Understanding, 78, no. 1, pages 99-118, 2000.
4. M. Jogan, E. Žagar, A. Leonardis. »Karhunen-Loéve expansion of a set of rotated templates«. IEEE trans. image process., July 2003, vol. 12, no. 7, str. 817-825.
5. S. Fidler, D. Skočaj, A. Leonardis. »Combining reconstructive and discriminative subspace methods for robust classification and regression by subsampling«. IEEE trans. pattern anal. mach. intell.. Mar. 2006, vol. 28, no. 3, str. 337-350.
Celotna bibliografija je dostopna na SICRISu: http://sicris.izum.si/search/rsr.aspx?lang=slv&,id=5591.