2018-00873 - Concept Detection in Images from Cultural Heritage databases
Le descriptif de l’offre ci-dessous est en Anglais

Contrat renouvelable : Oui

Niveau de diplôme exigé : Thèse ou équivalent

Fonction : Post-Doctorant

Niveau d'expérience souhaité : Jusqu'à 3 ans

A propos du centre ou de la direction fonctionnelle

The Inria Rennes - Bretagne Atlantique Centre is one of Inria's eight centres and has more than thirty research teams. The Inria Center is a major and recognized player in the field of digital sciences. It is at the heart of a rich R&D and innovation ecosystem: highly innovative PMEs, large industrial groups, competitiveness clusters, research and higher education players, laboratories of excellence, technological research institute, etc.

Contexte et atouts du poste

The Linkmedia team (https://www-linkmedia.irisa.fr/) in Rennes has one open post-doctoral position.
The proposed work is related to a European interdisciplinary R&D project (READ-IT) involving ICT scholars and Human and Social Sciences (HSS) scholars. The aim of Human and Social Sciences (HSS) researchers is to exploit rich 'human archives' in multiple media and languages depicting reading experiences, e.g., the practical circumstances of, the facts and effects related to reading. Visual representations of reading, such as photos, paintings, drawings, engravings, or sculptures portraying readers, constitute a particularly valuable source for the history of reading that has been neglected until now, partly because of the lack of automated visual content analysis tools to explore at scale such resources.

Mission confiée

The goal of this work is to facilitate access to an enriched dataset of visual sources, in order to study representations of reading from past and present in a computer assisted way, building on a close collaboration between scholars in reading studies and scientists in multimedia and computer vision.


The final goal would be for example to be able to evaluate in a systematic way the qualities of visual sources (e.g., distinguish between stereotypical vs. more realistic visualizations), get new insights into changing attitudes (individual or group reading), study the appearance of new readers over time (women, lower classes, children, ethnic minorities), or analyze changes in mass consumption of genres (bibles, novels, newspapers, comic strips) and reading devices (books, computers ipads, e-readers).
Investigating these key questions requires the development of high-level visual concept recognition tools for establishing comparisons (e.g., search for similar visual representations) and for automatic content analysis (recognition of visual concepts).

Principales activités

The goal is to analyze images from ‘human archives’: (i) to identify how many persons are involved, their status, their position, if they are sitting or not, who is reading, what is the environment, (ii) also to discriminate between reading and non-reading situations, meaning trying to detect if there is an interaction or an engagement between a person and a text.

This calls for adaptable technology to detect domain-specific concepts in images, leveraging state-of-the-art deep learning approaches and domain adaptation to cope with the limited amount of available annotated data.
Two key challenges arise: learning from very few examples, a frequent situation in HSS, and constantly learning as new data is being ingested.


Detailed post-doc subject can be found on the following web link: https://www-linkmedia.irisa.fr/files/2018/06/TrameOffreRecrutementPostDoc-2018.pdf

Compétences

Requirements
- PhD in machine learning, computer vision, or image analysis
- Relevant research skills and experience in computer vision, machine learning / deep neural networks
- Programming skills (e.g. Python or C++)
- Good English communication skills (oral and written)

Requirement due to funding:
- Maximum 3 years of experience after thesis defense
- An international experience in research is required (during or after Doctorate).
- Candidates must not have supported their thesis at UBL and not previously worked in the IRISA research unit.

Applying

Please send the following documents before 8 july 2018:

•    Short Curriculum Vitae and a covering letter showing your interest and especially addressing your professional project
•    A list of your major works (2 pages max.) : scientific publications, patents and others scientific productions
•    Letters of recommendation (not mandatory, but appreciated)
•    A copy of your PhD diploma  

by email (mentioning "Read-it postdoc position" in the email subject) to:
Ewa KIJAK (ewa.kijak@irisa.fr) and Brigitte OUVRY-VIAL (brigitte.ouvry-vial@univ-lemans.fr) with copy to UBL (recherche@u-bretagneloire.fr)

Avantages sociaux

 

  • Subsidised catering service
  • Partially-reimbursed public transport

Rémunération

monthly gross salary amounting to 2653 euros