Course 2019-2020 a.y.

20600 - DEEP LEARNING FOR COMPUTER VISION

Department of Marketing

Course taught in English
Go to class group/s: 31
DSBA (6 credits - I sem. - OBCUR  |  2 credits SECS-P/10  |  4 credits SECS-P/08)
Course Director:
GAIA RUBERA

Classes: 31 (I sem.)
Instructors:
Class 31: GAIA RUBERA


Suggested background knowledge

Good knowledge of both Python and R.

Mission & Content Summary

MISSION

Data is everywhere and it can be of any kind. Students enrolled in the Business Analytics track of the MSc in Data Science and Business Analytics have learned how to analyze numerical and textual data. This course augments their knowledge by teaching them how to deal with and analyze images. Images play a major role in today's communication, whether is corporate and official one or just a series of posts on Social Media. The course introduces the building blocks of modern Computer Vision algorithms by providing the students with a set of practical skills which can be applied in both the industry and academia.

CONTENT SUMMARY

  • Theoretical understanding of different deep learning model architectures.
  • Software libraries for Computer Vision.
  • Image classification.
  • Object detection.
  • Image segmentation.
  • Versioning software systems like git.

Intended Learning Outcomes (ILO)

KNOWLEDGE AND UNDERSTANDING

At the end of the course student will be able to...
  • Describe the main characteristics of a Computer Vision dedicated neural network.
  • Recognize limitations of a given deep learning model.
  • Select the appropriate model for a given Computer Vision task.

APPLYING KNOWLEDGE AND UNDERSTANDING

At the end of the course student will be able to...
  • Apply Computer Vision methods to solve real-world problems.
  • Analyze, compare, assess the performance, and select the appropriate algorithm.
  • Use relevant software libraries such as OpenCV, Tensorflow, Keras.
  • Design the appropriate workflow to correctly organize, implement, and later manage a Computer Vision project.
  • Develop a proper local and remote repository hosted on dedicated servers like Github.

Teaching methods

  • Face-to-face lectures
  • Guest speaker's talks (in class or in distance)
  • Exercises (exercises, database, software etc.)
  • Group assignments

DETAILS

In addition to face-to-face lectures, the learning experience of this course includes state-of-the-art model applications, hackaton competitions among enrolled students, and interactions with guest speakers from companies. Students are encouraged to actively participate to classes and interact with guest speakers in order to use their communication and interpersonal skills.


Assessment methods

  Continuous assessment Partial exams General exam
  • Written individual exam (traditional/online)
     x
  • Group assignment (report, exercise, presentation, project work etc.)
     x

ATTENDING STUDENTS

  • One final group project representing 70% of final grade.
  • One final written exam representing 30% of final grade.

NOT ATTENDING STUDENTS

  • One final individual project representing 40% of final grade.
  • One final written exam representing 60% of final grade.

Teaching materials


ATTENDING AND NOT ATTENDING STUDENTS

  • Slides provided by the instructors.
  • Francois CHOLLET (edition by Manning Publications): Deep Learning with R.
  • Francois CHOLLET (edition by Manning Publications): Deep Learning with Python.
Last change 03/05/2019 13:19