Offre en lien avec l’Action/le Réseau : – — –/Doctorants

Laboratoire/Entreprise : Hubert Curien Laboratory – UMR 5516
Durée : 1 year (possibly 2 y
Contact : alain.tremeau@univ-st-etienne.fr
Date limite de publication : 2023-02-01

Contexte :
The Image Science and Computer Vision team of Hubert Curien Laboratory (https://laboratoirehubertcurien.univ-st-etienne.fr/en/index.html) is looking for talents with skills and experience in video analysis and scene understanding for one Postdoc position (1 or 2 years) from January (or February) 2023.

This position is offered in the frame of the Horizon Europe project PREMIERE (Performing arts in a new era: AI and XR tools for better understanding, preservation, enjoyment and accessibility). The recruited candidate will be responsible for conducting research within the PREMIERE project and for participating in the related Work Package activities.

This project involves three permanent researchers from the laboratory and another postdoc who will collaborate with you and to the different work packages.

See https://premiere-project.eu/

Sujet :
The candidate will carry out research on developing tools and methods for 2D human pose estimation. Several topics will be investigated:
– Design of evaluation metrics for comparison of 2D human pose estimation neural network architectures (as example see https://github.com/open-mmlab/mmpose) and for improvement of existing neural network frameworks / loss functions.
– Selection of the “best” 2D human pose estimation methods (as the ones very recently surveyed in https://arxiv.org/abs/2204.07370) for a set of 2D videos from archives of performances/dances.
– Design of a strategy to transpose a human skeleton model to another model (with different keypoint and segment attributes) or to retrain a pose estimation model from another skeleton-based model than the one used for the pretraining.
– Design a framework to evaluate the quality of archives in terms of body pose estimation, hand pose estimation, face pose estimation, emotion recognition, etc.

Profil du candidat :
The ideal candidate for this Postdoc position has a PhD’s degree in deep learning and computer vision, with excellent academic records and research experience. A strong background in machine learning, artificial intelligence, data mining, is requested. Excellent programming skills, analytical problem solving and organizational abilities are required. Prior practical experience in deep neural networks is mandatory.
We are looking for a passionate researcher who loves challenges and who is highly qualified and engaged to study, design, develop and validate video analysis methods based on deep learning approaches.
Key qualifications:
– Strong experience in Python, including popular science libraries;
– Experience with one or more deep learning framework such as Tensorflow, PyTorch or Keras;
– Experience in research and/or development for action recognition, or human body pose and motion estimation, multi-objects tracking, activity recognition, scene understanding, etc.;
– Expert knowledge of the principles, algorithms, and tools of deep learning, including models, training strategies, training dataset building, loss functions, quality metrics, etc.;
– Excellent communication and collaboration skills;
– Excellent problem solving and analytical thinking skills;
– Excellent math skills: linear algebra, computational science and optimization;
– Creativity and curiosity for solving highly complex problems.

Formation et compétences requises :
PhD candidates expecting to finalize their degree in the coming months are also welcome to apply.
Application

Applications should be sent by email to Prof. Alain Trémeau (alain.tremeau@univ-st-etienne.fr) and should include the following documents.
1. A cover letter stating the purpose of the application and a brief statement of why you believe that your background and goals are well-matched with the goals of this position
2. A Curriculum Vitae that includes at least a list of previous degrees, dates, and institution, transcripts for higher-education studies until most recent available
3. Copies of previous transcripts and degree certificates
4. A summary (1-2 pages) of the PhD’s and/or master’s thesis
5. A copy of previous publications and software samples, if any, and
6. Reference letters or contact information for reference persons

List of qualifications (eg. in Software development using Python, in development frameworks for machine learning such as TensorFlow/Keras and PyTorch, or for real-time image processing like OpenCV) and other documents that the applicant wishes to refer to should be enclosed with the application.

Very good English communication skills and reporting, autonomy and curiosity, ability to work in a highly collaborative environment, sense of initiative and rigour will be greatly appreciated.

Adresse d’emploi :
Place of Work: Laboratory Hubert Curien, University Jean Monnet, Saint-Etienne, France