Course guide

230726 - CVDL - Computer Vision with Deep Learning

Last modified: 24/05/2024

Unit in charge: Barcelona School of Telecommunications Engineering
Teaching unit: 739 - TSC - Department of Signal Theory and Communications.

Degree: MASTER'S DEGREE IN TELECOMMUNICATIONS ENGINEERING (Syllabus 2013). (Optional subject).

MASTER'S DEGREE IN ADVANCED TELECOMMUNICATION TECHNOLOGIES (Syllabus 2019). (Optional
subject).

Academic year: 2024 ECTS Credits: 5.0 Languages: English

LECTURER

Coordinating lecturer: Consultar aquí / See here:

https://telecos.upc.edu/ca/estudis/curs-actual/professorat-responsables-coordinadors/respon
sables-assignatura

Others: Consultar aquí / See here:

https://telecos.upc.edu/ca/estudis/curs-actual/professorat-responsables-coordinadors/profess
orat-assignat-idioma

PRIOR SKILLS

Important: You should have the following previous knowledge to follow the course:
- Image processing: pixels, color spaces, histograms, frequency domain representation
- Digital signal processing: linear filters, convolution
- Vector and matrix algebra

Notions of python are useful, but these are easily obtained during the course.

DEGREE COMPETENCES TO WHICH THE SUBJECT CONTRIBUTES

Specific:
1. Ability to apply information theory methods, adaptive modulation and channel coding, as well as advanced techniques of digital
signal processing to communication and audiovisual systems.
2. Ability to integrate Telecommunication Engineering technologies and systems, as a generalist, and in broader and multidisciplinary
contexts, such as bioengineering, photovoltaic conversion, nanotechnology and telemedicine.

Transversal:
3. TEAMWORK: Being able to work in an interdisciplinary team, whether as a member or as a leader, with the aim of contributing to
projects pragmatically and responsibly and making commitments in view of the resources that are available.

4. EFFECTIVE USE OF INFORMATION RESOURCES: Managing the acquisition, structuring, analysis and display of data and information
in the chosen area of specialisation and critically assessing the results obtained.

5. FOREIGN LANGUAGE: Achieving a level of spoken and written proficiency in a foreign language, preferably English, that meets the
needs of the profession and the labour market.

Date: 30/05/2024 Page: 1 / 4

TEACHING METHODOLOGY

- Lectures
- Practical work
- Individual work (distance)
- Exercises
- Mid and final term exams

LEARNING OBJECTIVES OF THE SUBJECT

Learning objectives of the subject:

The aim of this course is to provide an overview of concepts and applications of computer vision, with both classic and Deep Learning
methods. We will introduce low level techniques such as feature extraction and matching, edge detection, cameras and projection
models and optical flow; mid-level topics such as video segmentation and feature tracking; high level methods such as object
tracking. Then, examples of application will be shown, such as face and object recognition.

Learning results of the subject:

- Ability to understand and use techniques for image and video analysis: feature extraction, video segmentation, stereo, object
detection.
- Ability to use computer vision algorithms to implement high-level applications.

STUDY LOAD

Type Hours Percentage

Self study 86,0 68.80

Hours large group 39,0 31.20

Total learning time: 125 h

CONTENTS

1. Introduction

Description:
- Motivation, types of problems in CV
- Image formation, perception, 3D sensors

Full-or-part-time: 7h
Theory classes: 3h
Self study : 4h

2. Image Structure

Description:
- Color, texture, filtering, and contours
- Detection and representation of interesting points and 'blobs'
- Modeling: RANSAC, Hough transform
- Saliency maps

Full-or-part-time: 29h
Theory classes: 9h
Guided activities: 4h
Self study : 16h

Date: 30/05/2024 Page: 2 / 4

 3. Stereo and 3D applications

Description:
- Single-camera geometry, camera calibration
- Epipolar geometry, homography
- Camera pose estimation and sensor registration using deep learning

Full-or-part-time: 30h
Theory classes: 9h
Guided activities: 4h
Self study : 17h

4. Video tracking

Description:
- Optical flow: Lucas-Kanade, Shi-Tomasi, Deep Learning methods
- Bayesian tracking: Kalman, Particle filters
- Deep Learning tracking methods

Full-or-part-time: 29h
Theory classes: 9h
Guided activities: 4h
Self study : 16h

5. Detection and recognition

Description:
- Introduction to visual recognition. Review of machine learning Deep learning and convolutional neural networks
- Image classification: Bag of words model. Image classification using CNNs
- Object detection: Sliding windows and local features. Object detection using CNNs
- Object segmentation: Semantic segmentation. Instance segmentation

Full-or-part-time: 30h
Theory classes: 9h
Guided activities: 4h
Self study : 17h

ACTIVITIES

EXERCISES

Description:
- Detecting contours and modeling shapes: Canny, Hough, Ransac, DL
- Finding correspondences between images: Harris, SIFT
- Fundamental matrix estimation
- Application of homography: panorama creation
- Object detection & recognition

Full-or-part-time: 6h
Self study: 6h

Date: 30/05/2024 Page: 3 / 4

 EXTENDED ANSWER TEST

Description:
Mid-term examination.

Full-or-part-time: 2h
Theory classes: 2h

EXTENDED ANSWER TEST

Description:
Second term examination

Full-or-part-time: 2h
Theory classes: 2h

GRADING SYSTEM

First-term examination: 40%

Second term examination: 40%
Laboratory/Exercises assessments: 20%

BIBLIOGRAPHY

Basic:
- Szeliski, R. Computer vision: algorithms and applications [on line]. London: Springer, 2011 [Consultation: 20/10/2014]. Available
on: http://site.ebrary.com/lib/upcatalunya/docDetail.action?docID=10421311. ISBN 9781848829350.
- Forsyth, D.A.; Ponce, J. Computer vision: a modern approach [on line]. 2nd ed. Boston, Mass.: Pearson Education, 2012
[Consultation: 09/09/2020]. Available on: https://ebookcentral.proquest.com/lib/upcatalunya-ebooks/detail.action?docID=5173504.
ISBN 9780273764144.

Complementary:
- Hartley, R.; Zisserman, A. Multiple view geometry in computer vision. 2nd ed. Cambridge: Cambridge University Press, 2003. ISBN
0521540518.
- Wang, Y.; Ostermann, J.; Zhang, Y.-Q. Video processing and communications. Upper Saddle River: Prentice Hall, 2002. ISBN
9788131733646.
- Hanjalic, A. Content-based analysis of digital video [on line]. Boston: Kluwer Academic, 2004 [Consultation: 29/07/2013]. Available
on: http://link.springer.com/book/10.1007/b106003/page/1. ISBN 978-1402081149.

RESOURCES

Other resources:
Google Colab

Date: 30/05/2024 Page: 4 / 4