In this course we will analyze the paradigm of automatic object recognition from a Computer Vision and Machine Learning points of view. We will review past and recent challenges in object recognition, such as multi-modal, multi-part, multi-scale, multi-view, multi-class, multi-label, and large scale object recognition. We will also review current trends for a particular and complex kind of objects: people in visual data'. We will deal with the problem of human pose recovery and automatic behavior analysis, describing potential applications as well as future lines of research in the field. Mail:
Teachers
Person in charge
Simone Balocco (
)
Others
Sergio Escalera (
)
Weekly hours
Theory
1.5
Problems
0
Laboratory
1
Guided learning
0
Autonomous learning
0
Competences
Generic Technical Competences
Generic
CG2 - Capability to lead, plan and supervise multidisciplinary teams.
Technical Competences of each Specialization
Academic
CEA6 - Capability to understand the basic operation principles of Computational Vision main techniques, and to know how to use in the environment of an intelligent system or service.
CEA14 - Capability to understand the advanced techniques of Vision, Perception and Robotics, and to know how to design, implement and apply these techniques in the development of intelligent applications, services or systems.
Professional
CEP6 - Capability to assimilate and integrate the changing economic, social and technological environment to the objectives and procedures of informatic work in intelligent systems.
CEP8 - Capability to respect the surrounding environment and design and develop sustainable intelligent systems.
Transversal Competences
Appropiate attitude towards work
CT5 - Capability to be motivated for professional development, to meet new challenges and for continuous improvement. Capability to work in situations with lack of information.
Basic
CB7 - Ability to integrate knowledges and handle the complexity of making judgments based on information which, being incomplete or limited, includes considerations on social and ethical responsibilities linked to the application of their knowledge and judgments.
Contents
Introduction to object and human recognition
Multi-modal object recognition
Multi-part object recognition
Multi-scale object recognition
Multi-view object recognition
Multi-class object recognition
Multi-label object recognition
Multi-ple data: large scale object recognition
Object Recognition in context: scene understanding and grammars
Human Pose Recovery
Human Behavior Analysis
Activities
Theoretical class
Theory
22.5
Problems
0
Laboratory
0
Guided learning
0
Autonomous learning
0
Practical sessions
Theory
0
Problems
0
Laboratory
15
Guided learning
0
Autonomous learning
0
Teaching methodology
T Each week it will be a 1.5h theoretical topic exposition class.
P Each week it will be a 1h practical session.
The rest of the course are devoted to autonomous lectures, programming, and studying.
Evaluation methodology
The course will follow a continuous evaluation consisting in four practical reports (PR) and two in-class presentations (PS). At the end of the course a test exam will be performed (TS). The final score (FS) will be computed as follows:
FS = 0.5 * PR + 0.3 * PS + 0.2 * TS
A minimum score of 3 over 10 points is required for each part PR, PS, and TS in order to compute the final score FS.
Bibliografy
Basic:
Computer Vision: A Modern Approach -
David A. Forsyth, Jean Ponce, ,
2002.
ISBN:
Computer Vision: Algorithms and Applications -
Richard Szeliski, ,
2010.
ISBN: http://szeliski.org/Book/
Complementary:
Human Behavior Analysis from Depth Maps -
Sergio Escalera, AMDO ,
2012.
ISBN:
Object detection with discriminatively trained part-based models -
P. Felzenszwalb, R. Girshick, D. McAllester, D. Ramanan, PAMI ,
2010, vol. 32, num. 9.
ISBN:
Multi-scale stacked sequential learning -
C. Gatta, E. Puertas, and O. Pujol, Pattern Recognition ,
2011, vol. 44, issue 10-11, pp. 2414-2426.
ISBN:
Multiple View Geometry Vision -
R. Hartley and A. Zisserman, ,
.
ISBN:
On the Decoding Process in Ternary Error-Correcting Output Codes -
Sergio Escalera, Oriol Pujol, and Petia Radeva, IEEE PAMI ,
2010, vol. 32, issue 1, pp. 120-134.
ISBN: 0162-8828 http://cataleg.upc.edu/record=b1000619~S1*cat
Subclass Problem-dependent Design of Error-Correcting Output Codes -
Sergio Escalera, David Tax, Oriol Pujol, Petia Radeva, and Robert Duin, IEEE Transactions in Pattern Analysis and Machine Intelligence ,
2008, vol. 30, issue 6, pp. 1041-1054.
ISBN:
An extensive experimental comparison of methods for multi-label learning -
Gjorgji Madjarov, Dragi Kocev, Author Vitae, Dejan Gjorgjevikj, Sao Deroski, Pattern Recognition ,
2012.
ISBN:
Sub-linear Indexing for Large Scale Object Recognitiom -
Stepan Obdrzalek and Jiri Matas, BMVC ,
2005.
ISBN:
80 million tiny images: a large dataset for non-parametric object and scene recognition -
A. Torralba, R. Fergus, W. T. Freeman, PAMI ,
2008.
ISBN:
The role of context in object recognition -
Oliva, A. Torralba, Trends in Cognitive Sciences ,
2007.
ISBN:
Objects in Context -
Rabinovich, A. Vedaldi, C. Galleguillos, E. Wiewiora and S. Belongie, ICCV ,
2007.
ISBN:
A Stochastic Grammar of Images -
S.C. Zhu and D. Mumford, Foundations and Trends in Computer Graphics and Vision ,
2006.
ISBN:
Articulated pose estimation with flexible mixtures-of-parts -
Y. Yang, D. Ramanan, Computer Vision and Pattern Recognition (CVPR) ,
2011, pp. 13851392.
ISBN:
Real-time American Sign Language Recognition using desk and wearable computer based video -
T. Starner, J. Weaver, and A. Pentland, IEEE TPAMI ,
1998, vol. 20, issue 1, pp. 1371-1375.
ISBN:
A tutorial on Hidden Markov Models and selected applications -
L. Rabiner, IEEE Speech recognition ,
1989, vol.2, pp.257-286.
ISBN:
A probabilistic framework for matching temporal trajectories: Condensation-based recognition of gestures and expressions -
M. Black and A. Jepson, LNCS ,
1998, vol. 1406, pp. 909-924.
ISBN:
Recognition and reproduction of gestures using a probabilistic framework combining PCA, ICA and HMM -
S. Calinon, A. Billard, ICML ,
2005.
ISBN: