Dynamic Vision for Actions

程一-计算所

Dynamic Vision for Actions

Application deadline: 15 June 2016.

Location: Singapore

Two Research Fellow positions are available at the Department of Electrical and Computer Engineering, National University of Singapore (NUS), to work under the project entitled Dynamic Vision for Actions. The broad objective of the project is to create a vision system that can move and act in the real world in an online manner (for both indoor and outdoor scenes).
One part of the project tackles the key competencies required by a vision-based autonomous system to move and function in the real-world which might contain multiple moving agents. Building on the recent progress on tracking and motion segmentation research, we will investigate the mid-level geometric interpretation in the face of multiple (possibly non-rigid or hierarchical) motions, as well as multi-object tracking. We also have to make sense of the 3D scene estimates for the robotic systems to act, for instance, in terms of some forms of scene layout navigable area, understand what are the objects and actions we see, etc.   
The other part of the project deals the key vision problems of labelling, assignment, clustering, and model selection encountered in the first part of this project but also in a more general setting. It aims to augment on the one hand the efficient move-making algorithms for solving clustering and assignment problems, and on the other to extend the modelling capability of MRF/CRF in labelling by “borrowing” features from the clustering and assignment formulations.  
The successful candidate should have a Ph.D. or equivalent degree, and he/she should have a strong background in the aforementioned areas, and in applied mathematics knowledge like large scale convex optimization and sparse representation, as well as proficiency in programming. Strong publication records in internationally recognized conferences or journals are preferred.
The positions are funded for 3 year, and are available from April onwards. Salary is competitive. Please send a detailed CV with publication list, a concise description of research interests and future plans, and academic transcripts to

Dr Cheong Loong Fah
Department of Electrical & Computer Engineering
Faculty of Engineering
4 Engineering Drive 3
Singapore 117576.
Email: eleclf@nus.edu.sg.

We will begin reviewing applications for the position immediately.