EUROPEAN COMMISSION Project Acronym : CATEGORIES
CALL IST-2001-IV-.21 Proposal Number : IST-2001-34067
Action Line IST-2001-IV.2.1 Date : 11.10.2001
INFORMATION SOCIETIES TECHNOLOGY
(IST)
PROGRAMME
CATEGORIES
Cognitive Vision: Automated Techniques for Effective Goal-based Object Reasoning in Intelligent Embedded Systems
Part C
Proposal number: IST-2001-34067
Call part identifier: IST-01-7-1A
Action Line : IST-2001-IV.2.1
Date: 11 October 2001
C2. CONTENT LIST
C2. CONTENT LIST 2
C3. Community added value and contribution to EC policies 3
C3.1 Transnational Approach 3
C3.2 European Policies and Relation to Standardisation 4
C4. Contribution to Community social objectives 5
C4.1 Quality of Life, Health and Safety 5
C4.2 Employment 5
C4.3 Market Potential 5
C5. Project management 7
C6. Description of the consortium 9
C7. Description of the participants 11
C7.1 ERCIM - European Research Consortium for Informatics and Mathematics 11
C7.2 INRIA – Institut National de Recherche en Informatique et en Automatique 13
C7.3 UoS - University of Sussex (COGS) 15
Company profile - www.cogs.susx.ac.uk 15
C 7.4 University of Amsterdam (UVA) 17
Company profile - www.uva.nl 17
ISIS group 17
Key Persons 17
Selected References 18
C 7.5 UBP - Université Blaise Pascal (LASMEA) 19
C 7.6 UCO - University of Córdoba 20
C7.7 University of Rennes 1 (Géosciences Rennes) 22
C7.8 TIMEAT SA 23
C8. Economic development and scientific and technological prospects 25
C9 Annex C1 – References 28
C3. Community added value and contribution to EC policies C3.1 Transnational Approach
Cognitive Vision relies on two distinct yet essential components: Knowledge and Computer Vision. The whole scientific challenge consists therefore in:
-
Gathering human expertise; Treating this information to obtain the essential human knowledge; Translating this expertise into computer language; Setting up corresponding databases; Designing intelligent systems able to perform human tasks autonomously and to learn from examples. This is all the more difficult as palynology remains a difficult science in which experts combine technology, experience and hand-crafted heuristics to identify and categorise pollen.
-
Designing a reliable Cognitive Vision system; Identifying specific objects and elements regardless of perturbations like noise, dirt etc.; Linking Computer Vision and Artificial Intelligence algorithms to a knowledge database for Automated Classification of objects and elements. This will be also ensured by CATEGORIES, especially when comes to translating human know-how into computer language and providing the system the means to learn from examples.
In order to address the first challenge, it is essential to gather the available expertise in each discipline. In this sense the best way to proceed is to rely upon European expertise rather than focusing individually on national competence. CATEGORIES sets out to develop a cognitive vision framework. The first domain studied in CATEGORIES are pollen and chitinozoans, the identification and categorising of which are made all the more difficult by its multiplicity and by the several varieties that so far only human eye and expertise have managed to differentiate.
As a consequence of this complexity no single partner can achieve the goal alone. Hence partners originating from four European countries (United Kingdom, Netherlands, Spain and France) have been chosen due to their complementary expertise. For example, among other expertise, while INRIA and the University of Amsterdam hold advanced knowledge in Computer Vision, University of Cordoba - REA and CNRS - Geosciences provide access to their expertise respectivelly in pollen and chitinozoans identification, expertise that COGS will exploit in the Knowledge and Learning module.
Related Projects
The consortium consists of experienced researchers with considerable and complimentary expertise in the areas required. All partners have successfully contributed to cooperative projects in the past, including many European projects. Cooperation with other related projects will be a priority in order to build synergies from which derive additional expertise. Project clustering is not only a way for the projects to exchange ideas and concepts, it is also an extensive way to share expertise and to benefit from other project achievements and failures in order to gain time and save man-power by going directly along the right track. The first contacts made under this initiative (ActIPret IST-2001-32184) have built up a very good working relationship and strong determination for active and harmonious collaboration. Specifically for the demonstrated application, the partners will contribute their knowledge and vision techniques developed in ASTHMA project (DGXII, 4th FP RTD).
Critical Mass
The partners would not be able to establish an equally competent consortium within their own countries. This is why the European Added Value of the Consortium allows the integration of expertise from the different areas involved. This is the only way to ensure a coherent overall view of the project’s objectives, to avoid interference or incompatibility between interdependent activities, and to achieve a high level of synergy between complementary efforts and previous work. CATEGORIES integrates together private companies and research institutions that provide all the needed fields of expertise to build a framework for automated categorisation from visual input. Most partners have expertise in cooperative projects and some have already cooperated, therefore effective integration of the team will be obtained quickly in order to reach the goals with great efficiency.
These elements put together illustrate the European nature of this project, not only by the composition of the consortium but essentially given the nature of research itself. Indeed, the issue of Cognitive Vision is currently narrowly addressed by independent national initiative. In this respect, with the new workprogramme 2002 and in particluar action line IV.2.1, the European Commission therefore appears as a natural partner to CATEGORIES.
Dostları ilə paylaş: | 
