Projects / Programmes
Biometric face recognition in ambient intelligence environments (BAMBI)
| Code |
Science |
Field |
Subfield |
| 2.06.02 |
Engineering sciences and technologies |
Systems and cybernetics |
System theory and control systems |
| Code |
Science |
Field |
| P175 |
Natural sciences and mathematics |
Informatics, systems theory |
| Code |
Science |
Field |
| 2.02 |
Engineering and Technology |
Electrical engineering, Electronic engineering, Information engineering |
biometrics, ambient intelligence environments, face recognition, software
Researchers (1)
| no. |
Code |
Name and surname |
Research area |
Role |
Period |
No. of publicationsNo. of publications |
| 1. |
28458 |
PhD Vitomir Štruc |
Systems and cybernetics |
Head |
2011 - 2013 |
361 |
Organisations (1)
Abstract
The development of information technology and ever more powerful computers, the emergence of tablets and the wide spread availability of mobile devices allow users to access different databases, services and other e-applications with an increased ease. However, these new possibilities have also triggered a need for secure authentication schemes better suited to meet the demands of society than the traditional token- or knowledge-based means of identification. The current trend in meeting these needs is to design biometric security systems which exploit the physiological or behavioral characteristics of an individual to establish identity. While the primary use most often associated with biometric recognition systems is still within security application, biometrics is slowly moving into other deployment areas as well. An important example of these areas also includes ambient intelligence applications.
The field of ambient intelligence strives toward more human-centric technology, where people instead of technology are at the center of attention. By definition an ambient intelligence environment is capable of: detecting (or sensing) the presence of a known person, adapting to the sensed person, and interacting with the person, while running in the background undetectable to the human eye. The idea of ambient intelligence has gained a lot of popularity in recent years, which is also evidenced by the actions of the European Commission, which supported the idea of ambient intelligence environments with the establishment of the advisory group IS-TAG. The group composed a report entitled “Scenarios for Ambient Intelligence in 2010”, which identifies ambient intelligence applications as one of the priority research fields in the EU.
In the scope of the proposed two-year applied postdoctoral project BAMBI the postdoctoral candidate will develop a prototype face recognition system for ambient intelligence environments. With the developed prototype it will be possible to ensure one of the fundamental building blocks of ambient intelligence environments, namely, the ability to detect and recognize users known to the environment.
During his work on the postdoctoral project the project leader will devote his attention to solving the problems of the existing face recognition techniques encountered when deployed in uncontrolled conditions such as those met in ambient intelligence environments. These problems typically relate to the variability of the external illumination conditions, ageing of the user base, pose variations and the interaction of these influential factors. To tackle these problems the postdoctoral candidate will develop novel procedures for robust face detection and registration, novel preprocessing techniques, and new statistical modeling approaches that are employable in ambient intelligence environments. The developed prototype face recognition system, which will present the main result of the project, will integrate all of the developed procedures and will later be evaluated on publicly available databases of face images.
The experience of the research group of the Laboratory of Artificial Perception, Systems and Cybernetics, the expertise of the external industrial partner, who will (co)finance the project, and the insight into the structure, operation and characteristics of biometric recognition systems that the postdoctoral candidate has obtained during the work on his PhD thesis, will all help to achieve the project goals.
Significance for science
We believe that the work, conducted in the scope of the postdoctoral project BAMBI, contributed significantly to the development of the field of pattern recognition, biometrics and facial recognition on the global scale. The value and importance of our work was demonstrated through peer-reviewed publications in established (SCI-indexed) journals, leading international conferences and last but not least at the international competition on face recognition, where the recognition approach developed based on the findings of the BAMBI project achieved the best recognition performance. Amon the more important contributions of the BAMBI project we would like to highlight: • The development of non-parametric and composite (hybrid) score normalization techniques, which are capable of significantly improving the robustness of facial recognition systems with respect to external factors affecting the facial appearance, such as ageing, lighting changes and alike. The techniques, which were initially presented at international conferences, were recently extended and are now waiting for publication at the »IET Biometrics« journal. The proposed techniques are important not only to the field of facial recognition, but to related fields such as speaker, fingerprint of palm print recognition as well. Furthermore, they are also applicable to any two-class recognition problem, where the problem is initially casted as a multi-class problem. • The implementation of reference algorithms and tools used in the field of facial recognition. The Matlab toolboxes INFace and PhD, which were either written or up-dated in the scope of the BAMBI project, offer implementations of state-of-the-art recognition and preprocessing techniques and therefore, allow researchers to focus on their contributions instead of re-implementing known, existing techniques. This in turn leads to a faster development of the field and is extremely important from the scientific perspective as well. The popularity of our toolboxes is evidenced by the number of downloads, which already reached several thousand for both toolboxes. • The introduction of the simplified and patch-wise simplified probabilistic linear discriminant analysis, which are capable of extracting highly discriminant features from facial images and have demonstrated to be suitable for the implementation of robust face recognition systems typically needed in ambient intelligence environments. The main characteristics of the two techniques include low-RAM requirements during run-time, the possibility of incremental template up-dating and high recognition performance. Since very similar procedures are also used in other areas, such as speaker recognition, the two techniques have a direct impact on this and related fields as well. • The importance of our work for the development of science was also demonstrated with the application of some of the developed techniques on other areas, such as the area of emotion recognition and 3D face recognition. The work in this field let to several publications, which show the interdisciplinary of our work. A couple of more important references is given below: o DOBRIŠEK, Simon, GAJŠEK, Rok, MIHELIČ, France, PAVEŠIĆ, Nikola, ŠTRUC, Vitomir. Towards efficient multi-modal emotion recognition. International journal of advanced robotic systems, ISSN 1729-8814, 2013, vol. 10, no. 53, str. 1-10, ilustr. doi: 10.5772/54002. [COBISS.SI-ID 9608276] in o KRIŽAJ, Janez, ŠTRUC, Vitomir, DOBRIŠEK, Simon. Combining 3D face representations using region covariance descriptors and statistical models. V: 10th IEEE International Conference on Automatic Face and Gesture Recognition, Shanghai, China, April 22-36, 2013. FG 2013. [Piscataway]: Institute of Electrical and Electronics Engineers: = IEEE, cop. 2013, str. 1-7, ilustr. [COBISS.SI-ID 9821012]
Significance for the country
When presenting the impact of the post-doctoral project BAMBI on the development of Slovenia several impact-levels can be pointed out: • Impact on the co-financing institution: Alpineon d.o.o., who was co-funding the project gained new knowledge, competences and expertise through various results of the BAMBI project (e.g., deliverables, publications and the prototype system). The obtained expertise can be put to practice in other fields Alpineon is active, used with other industrial problems or adopted for the development of facial recognition technology. We believe that through the BAMBI project Alpineon gained important competences, which will help it with its efforts on the market. • Impact on the institution, where the project was conducted (FE UL): The BAMBI project helped the Faculty of Electrical Engineering of the University of Ljubljana to stay in touch with most recent developments in the area of face recognition and biometrics in general and established itself further as a serious research institution working in this field. The work on the BAMBI project, hence, helped to increase the visibility of the faculty in the relevant research community. • Impact on the economic sector: Based on the expertise the project leader, ass. prof. Vitomir Štruc, obtained through his work on the postdoctoral project BAMBI, he established a sole proprietorship, VitPat, Vitomir Štruc, s.p., where he is currently trying to commercialize his expertise. • Impact on research activities: Most of the procedures developed in the scope of the BAMBI project have already been published in one form or another. Since all procedures are in the public domain, we believe anyone interested in tackling the problem of face recognition in Slovenia has a good starting point. Furthermore, the developed INFace and PhD toolboxes offer a good starting point for engaging into research on face recognition. • Impact on the promotion of Slovenia at the international level: The BAMBI project leader managed to win the bid for the organization of the 11th IEEE International Conference on Automatic Face and Gesture Recognition (AFGR). This achievement was made possible in parts due to the references obtained in the scope of the BAMBI project. The AFGR conference series is one of the leading conference series in the world in the field of computer vision and pattern recognition and will be held in 2015 in Slovenia. The right to organize the conference represents a big success that will have a direct impact on Slovenia’s economy and will greatly improve Slovenia’s visibility within the international research community.
Most important scientific results
Annual report
2011,
2012,
final report,
complete report on dLib.si
Most important socioeconomically and culturally relevant results
Annual report
2011,
2012,
final report,
complete report on dLib.si
