Dr Panraphee Raphiphan (graduated 2015) See picture of Panraphee above.
After completing her PhD with FIT Panraphee returned to Thailand where she works in the media as an Entrepreneur and as a University lecturer. Her media work stretches across television, radio and online, channels focused on technology in everyday life. In addition, she produces her own Tech show called iT 24 Hours (iT24Hrs) on nationwide TV channels and on radio.
Panraphee’s research for her PhD was focused on a context-aware traffic congestion estimation framework to overcome missing sensory data in Bangkok. Traffic congestion is the cause of pollution and economic loss. The real time traffic state report can alleviate this
problem by assisting drivers for route planning and choosing unblocked roads. More complete traffic information could lead to more accurate route planning. To ensure the continuity of reported traffic information, she propose the CATE framework that can estimates traffic information even though sensory data is unavailable and provide the guideline to improve existing traffic information system.
Dr Nitin Mahadeo (graduated 2015)
Nitin is now a working in the United States as a Research Fellow in Radiation Oncology Brigham and Women's Hospital, Boston, and before that he was a Postdoctoral fellow at Harvard Medical School. Nitin's PhD study involved investigating different techniques for improving the performance of iris recognition systems in less constrained environments using a video-based approach. Firstly, a new model for the extraction of eye images from NIR face videos is presented. Secondly, an optimal frame selection technique for enhancing the performance of video-based iris recognition systems is proposed. Thirdly, a robust iris segmentation system designed primarily for eye images captured in less constrained imaging conditions is presented. Fourthly, a model for automatically predicting iris segmentation errors in an iris recognition system is proposed. Finally, a method which takes advantage of the availability of multiple frames to build an optimized iris code is developed. Our results and experiments suggest that incorporation of the above methods in traditional iris recognition systems will be useful for the adoption of this technology by a larger community.