Computer Vision

Objectives

Computer Vision is called to revolutionise industrial automation and robotics as well as multimedia computing. Among the most important and powerful sensors both in automated factory environments and in non-industrial robotics applications are vision systems with cameras. This is also true for interactive installations and other multimedia applications. The aim of this course is to provide an introduction to computer vision and a solid background on image processing techniques with special emphasis on applications of interest in the fields of mechatronics and multimedia engineering. The course should provide the students with:

• Knowledge about the main techniques and tools to develop or assemble computer vision systems
• Ability to evaluate applications in the computer vision/image processing domain
• Ability to implement simple computer vision solutions in a laboratory environment
• Ability to independently develop a course project

Learning outcomes

• Analyse, design and solve event-driven programs in graphical environments with or without control. (96)
• Know the principles and techniques of image recognition and processing, and use them in industrial applications. (98)
• Analyse critically the results obtained. (94)
• Solves problems and situations typical of professional activity with entrepreneurial and innovative attitudes. (99)

Competencies

General skills

• Be prepared to overcome adversity in professional activity and learn from mistakes in order to integrate knowledge and enhance one's preparation. 

Specific skills

• Use dynamic system modelling tools and simulation techniques. Understand and apply the properties of sensors, actuators and signal conditioners, in order to program programmable robots, numerical control and robots to develop complex robotic systems that improve processes and the final product.

Basic skills

• Students have the ability to gather and interpret relevant data (usually within their field of study) in order to make judgements that include reflection on relevant social, scientific and ethical issues.

Core skills

• Project the values of entrepreneurship and innovation in one's academic and professional career, through contact with a variety of practical contexts and motivation for professional development.

Content

1. The human visual system and computer vision systems
2. Digital image fundamentals
   • Image representation
   • Colour image processing
3. Image enhancement
   • Spatial domain methods
   • Frequency domain methods
4. Image understanding
   • Feature extraction
   • Pattern recognition
   • Introduction to deep learning

Evaluation

The evaluation of the course follows a continuous assessment methodology through the presentation of two practical assignments, and a course project. A validation test will take place at the end of the term to ensure that each student has individually achieved the main objectives of the course. During the last four weeks of the course, the students will work on a course project, which will be more ambitious in scope than the practical assignments as it will tackle a real challenge. To develop this project, the students will need to study and deepen into some concepts in an autonomous way. The team members must expose and report regularly on the status of the project. At the end of the course, the teams will publicly defend their project and will deliver a final report.

Final grade = 0.15·A1 + 0.2·A2 + 0.4·P + 0.25·VT(*)

• A1: Assignment 1 (will include self-assessment and between-peers assessment)
• A2: Assignment 2 (will include self-assessment and between-peers assessment)
• P: Final Project (individual and collective evaluation)
• VT: Validation test

(*) A minimum score of 4 will be required in the VT to pass the course.

The students who do not pass the course can sit a final exam (E), in which case, the course grade will be calculated as follows:

Final grade = 0.3·C + 0.3·P + 0.4·E (**)

with

C = 0.4·A1 + 0.6·A2

(**) In order to be able to sit the final Exam (E), the students must have submitted all the course assignments during the course.

Methodology

The course combines master classes and a project based learning (PBL) methodology. The instructor will present the key concepts of each topic during the lectures, which should be consolidated both individually and cooperatively by the students. Such lectures will include hands-on practical exercises. There will also be specific sessions in which the students will work on practical assignments. These assignments will be subsequently delivered as part of a continuous evaluation framework. At the end of the course, a group project will be defined and executed. The project aims to develop technical and interpersonal skills, as well as individual responsibility, while providing a highly motivating context that will allow you to address your specific interests in the field. During the lectures, the working environment will be Matlab®.

Bibliography

Bibliography

• Davies, E.R. (2017). Computer Vision: Principles, Algorithms, Applications, Learning (5 ed.). Academic Press.
• Gonzalez, R.C., and Woods, R.E. (2006). Digital Image Processing (3 ed.). Prentice Hall.

Reading

Teachers will provide complementary bibliography and compulsory reading throughout the course via the Virtual Campus.