Computer Vision 1

Course manual 2019/2020

Course content

Digital cameras have become ubiquitous in the form of consumer cameras, webcams, mobile phones, and professional cameras. These cameras yield enormous streams of data and provide the means for communication, observation, and interaction. In this course, image understanding is addressed with the focus on core vision tasks of scene understanding and object recognition.

A broad range of techniques are studied on how computers can understand the visual world of humans including image formation and filtering, features (color and shape invariants, interest point detectors, descriptors, SIFT, HoG), visual information representation (vector space, statistical models, bag-of-words), learning and classification (nearest neighbor, kernel density estimation, SVM), dimension reduction (PCA, LDA and SVD), object detection and classification, object tracking (mean-shift, Kalman), and user interaction (active learning).

This year, different advanced applications in human behavior understanding are studied such as face and emotion recognition, human body analysis and affective computing. Further, we concentrate on object recognition in the field of computer vision. We discuss the data, tasks, and results of Pascal VOC and TRECVID, the leading benchmarks. In addition, we discuss the many derived community initiatives in creating annotations, baselines, and software for repeatable experiments.

Study materials

Literature

• Different chapters from: R. Szeliski, 'Computer Vision: Algorithms and Applications', 2010. The pdf is available for free: http://szeliski.org/Book/

Other

• The study material is given by the lecture slides, background papers, and relevant state of the art literature. In addition, seminar and lab exercises are provided.

Objectives

• To become familiar with both the theoretical and practical aspects of computer vision and to describe the foundations of image formation, measurement, and analysis.
• To learn physical-, statistical- and deep learning models as the foundation of computer vision and apply them to computer vision problems.
• To learn the fundamental aspects of different computer vision models and to explain their advantages and shortcomings
• Is able to implement computer vision programs and can evaluate/test these programs to understand the limitations of these vision algorithms.
• To learn how to develop the practical skills necessary to design, model, build, organize, and execute computer vision applications.

Teaching methods

Lectures, seminars and lab exercises.

Learning activities

Attendance

This programme does not have requirements concerning attendance (OER part B).

Assessment

Fraud and plagiarism

The 'Regulations governing fraud and plagiarism for UvA students' applies to this course. This will be monitored carefully. Upon suspicion of fraud or plagiarism the Examinations Board of the programme will be informed. For the 'Regulations governing fraud and plagiarism for UvA students' see: www.student.uva.nl

Course structure

Timetable

The schedule for this course is published on DataNose.

Additional information

Recommended prior knowledge: Programming experience and a general background in mathematics such as linear algebra, calculus and probability theory. A basic knowledge of image processing and computer vision.

Contact information

Coordinator

• prof. dr. T. Gevers