Home

Class for PhD in Engineering in Computer Science (http://www.dis.uniroma1.it/~dottoratoii/)

Teacher:

Prof. Ana C. Murillo

Informatics and Engineering Systems Dpt., Universidad de Zaragoza, Spain.

website: http://webdiis.unizar.es/~anacris

Dates, schedule and location: 14-18 March, 2016.

  • Mon 14/3 - 10.15-13.30 - Room A4
  • Tue 15/3 - 10.15-13.30 - Room A2
  • Wed 16/3 - 10.15-13.30 - Room A2
  • Thu 17/3 - 10.15-13.30 - Room A3
  • Fri 18/3 - 14.00-17.15 - Room A2

Description and Goals:

This course will review basic and widespread techniques for visual recognition, with special focus on specific techniques for recognition of human related concepts and activities: face and people detection and classification; analysis of human activity; recognition from mobile cameras (in particular egocentric cameras, i.e., mounted on a person or a mobile robot).

Lectures will combine description and discussion of computer vision techniques as well as demonstrations. There will be small exercises proposed and demonstrated in class about each topic. Students will have to combine at least two of the studied topics/techniques in these exercises into a final class project of their choice (among a set of application topics using the provided datasets). More details in the assignments description page.

Evaluation

This class will be evaluated from:

  • Small test during the last lecture about the concepts studied - 30%
  • Visual Recognition project (demo/video, presentation,poster) - 70%

Notes:

- Presentation will be given during last day lecture.

- Demo and Poster describing final version of the project can be turned in later (through hangout) if not finished during the course week.

Syllabus (tentative. more details in lectures description page)

  1. Image representation and visual classification
  2. Visual recognition of human features
  3. Visual recognition of human activities
  4. Visual recognition from mobile "egocentric" cameras (i.e, first person perspective, such as wearable and robot-mounted cameras)