Video Generation
using Deep Learning

Generative models have witnessed increasing attention due to their ability to model complex data distributions, which allows them to generate realistic images. While realistic video generation is the natural sequel, it is substantially more challenging w.r.t. complexity and computation, associated with the simultaneous modelling of appearance, as well as motion. Specifically, in inferring and modelling the distribution of human videos, generative models face three main challenges: (a) generating uncertain motion, (b) retaining human appearance throughout the generated video, as well as (c) modelling spatio-temporal consistency. Finding suitable representation learning methods, which are able to address these challenges, is critical to the final visual quality and plausibility of the rendered novel video sequences.

The Video Generation project will be executed at University of Warsaw, Faculty of Mathematics, Informatics, and Mechanics, University of Warsaw under the PI: Piotr Biliński, PhD.

Score of work:

• The team member will be working with the PI and will be fully involved with all matters of research.
• Research and development of novel computer vision and machine learning algorithms.
• Performing scientific experiments. Performing analysis of experiments and results.
• Writing timely reports on the research and publishing these as papers in leading top-tier conferences and journals.
• Performing occasional demonstrations to academic and industrial visitors.
• Contributing documented software to the group library.

Required professional qualifications:

• Good knowledge about the state-of-the-art in computer vision and machine learning.
• Good programming skills: Python, Numpy and PyTorch/ TensorFlow/ Keras/ Caffe/ similar deep learning framework.
• Good communication skills, including the ability to write for publication, present research proposals and results.
• Ability to work independently.
• Fluent English.

Stipend: up to 5000 PLN brutto brutto, depending on experience and time commitment, between June 4, 2024 and October 3, 2024, with the possibility to extend.

What else do we offer? Working on "hot" deep learning problems, paper co-authorship (in the case of publication), good athomsphere!

Application process: Please send the following documents to bilinski@mimuw.edu.pl by June 2, 2024 8 PM:

• Curriculum Vitae (Mandatory) and Cover Letter (Optional, but good to have) with statement: „Wyrażam zgodę na przetwarzanie moich danych osobowych dla potrzeb niezbędnych do realizacji procesu rekrutacji (zgodnie z Ustawą z dnia 29.08.1997 roku o Ochronie Danych Osobowych; tekst jednolity: Dz. U. 2016 r. poz. 922).”
• Information on how many hours a week you would like to dedicate to the project.
• Certification of being a student (if your are not a student of MIMUW).
• (Optional, but good to have) Link to papers.

Questions: Please don't hesitate to ask questions by sending an email to bilinski@mimuw.edu.pl.