Section: Highlights of the Year

Highlights of the Year

• The three-year FP7 STREP project Embodied Audition for Robots successfully terminated in December 2016. The project has addressed the problem of robot hearing, more precisely, the analysis of audio signals in complex environments: reverberant rooms, multiple users, and background noise. In collaboration with the project partners, PERCEPTION contributed to audio-source localization, audio-source separation, audio-visual alignment, and audio-visual disambiguation. The humanoid robot NAO has been used as a robotic platform and a new head (hardware and software) was developed: a stereoscopic camera pair, a spherical microphone array, and the associated synchronization, signal and image processing software modules.

• This year, PERCEPTION started a one year collaboration with the Digital Media and Communications R&D Center, Samsung Electronics (Seoul, Korea). The topic of this collaboration is multi-modal speaker localization and tracking (a central topic of the team) and is part of a strategic partnership between Inria and Samsung Electronics.

• Antoine Deleforge (former PhD student, PANAMA team), Florence Forbes (MISTIS team) and Radu Horaud received the 2016 Award for Outstanding Contributions in Neural Systems for their paper: “Acoustic Space Learning for Sound-source Separation and Localization on Binaural Manifolds," International Journal of Neural Systems, volume 25, number 1, 2015. The Award for Outstanding Contributions in Neural Systems established by World Scientific Publishing Co. in 2010, is awarded annually to the most innovative paper published in the previous volume/year of the International Journal of Neural Systems.

• Xavier Alameda-Pineda and his co-authors from the University of Trento received the Intel Best Scientific Paper Award (Track: Image, Speech, Signal and Video Processing) for their paper “Multi-Paced Dictionary Learning for Cross-Domain Retrieval and Recognition" presented at the 23rd IEEE International Conference on Pattern Recognition, Cancun, Mexico, December 2016 .

Best Paper Award: