Post-Doctoral Research Visit F/M High Performance Video Annotation in Sport

This job offer has expired

    Inria, the French national research institute for the digital sciences
    Computer science
    Recognised Researcher (R2)
    01/06/2021 00:00 - Europe/Brussels


The PerfAnalytics project, funded by ANR-PIA, involves several academic partners (Inria, INSEP, Univ. Grenoble Alpes, Univ. Poitiers, Univ. Aix-Marseille, Univ. Eiffel), as well as elite staff and athletes from different Olympic disciplines (Climbing, BMX Race, Gymnastics, Boxing and Wrestling).

The project investigates how computing systems can better assist professional coaches (and athletes) in their performance development. More precisely, its ambition is to capitalize on the high throughput capability offered by in-situ video analysis. However, because of tedious manual annotation pipelines, the in-situ exploitation of video content by coaches and athletes is still limited nowadays to a qualitative assessment through simple playback visualization.

In this project, we will design innovative workflows, interactions and interfaces to ease the process of in-situ video analysis of athlete performance in order to facilitate the optimal adjustment of strategy and training programs with respect to the athlete’s current state of performance. Notably, we will identify which level of human-machine partnership is recommended to bring the best of both worlds and combine semi-automatic rapid and precise video annotation and analysis, with efficient visual representation of performance.

This post-doctoral research project, within the Inria Loki group (http://loki.lille.inria.fr/), is about studying and designing the interaction techniques and paradigms that will successfully establish a strong partnership between expert users and learning algorithms, where both would improve their “skills” in order to reach a high level of efficiency for annotating videos. The researcher will study combinations of input and output methods for supporting this approach: manual or semi-manual annotations, validation/rejection of automatic annotations, real-time visualization of algorithms output, etc. Special care also has to be given to the adaptability of the proposed solution to different sports and users.

Current professional or experimental solutions (e.g. Dartfish, Meta-Video) for video annotation are fully manual and thus very constrained and limited in terms of performance. They can only be used offline, since in most sports, there is a high number of features to annotate plus many events per minute or second. It results in 1 minute of video taking about 3 minutes to be annotated (for e.g., Boxing). These video sequencers also rely on standard graphical user interfaces (windows, menus, buttons) which are clearly limiting fast and efficient annotation, because of the high number of different features and events that require to repeatedly trigger many different actions. But it also prevents the adaptation and the adaptability of the system to the particularities of different sports, and also to the needs and specificities of coaches’ practices. All in all, this situation partly explains the low adoption of video analysis by professional coaches and athletes.

Our goal is to design new interactive tools for video annotation based on a mixed initiative approach. The recent advances in machine learning allow robust identification of body parts from video. Especially with multiple viewpoints, most accessible solutions can predict human pose in terms of 3D location of joints centers. Partners of the project will study the adaptation of existing generic learning techniques to the requirements and context of the target disciplines. But this will also be a part of a larger interactive approach, in combination with manual scoring of actions from video.

The recruited person will work closely with with Stéphane Huot, Sylvain Malacria & Géry Casiez.
She/he will be instrumental in defining the research program for the workpackage we are involved in, and for coordination with the project patners.

  • Review related work and expert practices (e.g., field studies, interviews, etc.)
  • Design novel interaction techniques and visualization techniques for video annotation and performance analysis in sport with appropriate user-centered methodologies (e.g., participatory design)
  • Implement demonstrators and/or applications that can be used to test these designs
  • Design, run and analyze the results of controlled and/or field experiments
  • Contribute to project reports and write scientific papers
  • Contribute to coordination with project partners

More Information


The recruited person will join a dynamic team of international scientific experts in the field of Human-Computer Interaction (http://loki.lille.inria.fr/).

You will be contributing to a large-scale scientific project with a high potential impact as part of the specific research programme for high-performance sport in preparation for the 2024 Olympic Games in France.

We work in a stimulating and pleasant working environment (participation in transport (50%), on-site catering; teleworking; leave and special leave of absence (45 days), videoconference equipment, etc.).

Our agents benefit from quality training adapted to their needs and skills, whether technical, methodological or linguistic.

In addition to improving their technical skills, Inria offers our researchers the opportunity to develop their entrepreneurial skills by participating in awareness-raising events and training on the creation of start-ups (start-up horizon, intellectual property training, hackAthon ... https://www.inria.fr/fr/inria-startup-studio ).

Our administrative services assist new international agents with the various administrative procedures (visa, residence permit, social security, housing, bank, etc.)

Gross monthly salary (before taxes) : 2 653 €

Selection process

CV, application letter, list of publications, one or more letters of recommandation

Web site for additional job details

Offer Requirements

    FRENCH: Basic
    ENGLISH: Good

Specific Requirements

A PhD in the field of Human-Computer Interaction or Information Visualization is required.

A solid track record of publications in top-tier HCI or Information Visualization venues (such as ACM CHI, UIST, CSCW, InfoVis, ...) is expected, as well as a significant experience of design and implementation of interactive systems and GUIs.

A knowledge of, or at least a strong interest in, elite sports is a real asset.

Work Location(s)
1 position(s) available at France

EURAXESS offer ID: 613313
Posting organisation offer ID: 3357


The responsibility for the jobs published on this website, including the job description, lies entirely with the publishing institutions. The application is handled uniquely by the employer, who is also fully responsible for the recruitment and selection processes.


Please contact support@euraxess.org if you wish to download all jobs in XML.