Third course: (Tuesday February 2nd) will start at 1:30 pm as usual.

Until further notice all courses shall take place via zoom following this link. Stay tuned to the mailing list for possible changes in the zoom link, and for other important announcements related to the course.

All students planning to attend and/or validate this course should:

  1. Fill in this registration form
  2. Subscribe to the mailing list for important announcements related to the course

Course Description and Objectives

The minituriasation of sensors and the evolution of computational capabilities has led to the ubiquitous presence of images. However, the increasing demand for image-based content also requires sophisticated post-processing (filtering, restoration etc.), in order to ensure good quality results. At the heart of these post-processings are image models, which allow us to establish powerful and efficient algorithms. Deep learning is the latest addition to a long list of such image models. The aim of this course is to present recent techniques based on deep learning for the quality of images, and to compare them to pre-existing methods. We will explore many applications such as denoising, super-resolution, deblurring, texture synthesis and natural image generation. In each case we will present the strengths and limitations of the different techniques studied. In particular, we will present a critical analysis of the methods and show some of the pitfalls in which they sometimes fall.


The course may be given in English or French depending on the audience’s preferences.


  • Applied Mathematics (Linear Algebra, Numerical Analysis, Differential Calculus, Fourier Analysis) Programming (Python, Matlab)
  • Basic concepts of image processing, optimization and deep learning are useful but will be introduced in the course.


  • Individual reports of Practical Work (TPs).
  • Individual project with report and oral defense at the end of the course.

Teaching team

Alasdair Newson (MdC Telecom Paris)
Said Ladjal (MdC Telecom Paris)
Andrés Almansa (DR CNRS – MAP5 – Université Paris Descartes)

Organization of the course


  1. End-to-end deep learning and applications to super-resolution (S. Ladjal, 1 course + 1 TP)
  2. Inverse problems, variational, statistical and hybrid methods (A. Almansa, 3x course + TP)
  3. Generative models for texture synthesis (S. Ladjal, 1 course+TP)
  4. Variational Auto Encoders (A. Newson, 1 course + TP)
  5. Generative Adversarial Networks (A. Newson, 1 course + TP)

Detailed program…