I work in the Exploratory computer vision and pattern recognition group.
![]() |
![]() |
![]() |
![]() |
| I'm also working on fitting articulated models to video. This shows a still from the middle of a video constructed from the CMUMobo dataset, showing a person walking on a treadmill, from two different viewpoints. The coloured regions show the current fit of the model to the data. Click on the image for a video of automatic tracking (27MB AVI). | |
| 2001- | Research on visual person tracking in the PeopleVision project (above) | 1996- | Research on Biometric identification (face and fingerprint recognition). In particular I worked on fingerprint classification and distortion, and have written IBM's face detection, tracking and recognition system. This this is used as the visual side of the audio-visual speech recognition engine. |
| 1994-6 | Member of IBM's Handwriting Recognition Group, working on algorithms for on-line cursive handwriting recognition - getting computers to recognize handwriting written on a computer with an electronic stylus. |
| 1991-4 | Ph.D. (1994) Cambridge
University Engineering Department's
Speech, Vision and Robotics Group. My thesis `Off-line Handwriting Recognition Using Recurrent Neural Networks' was also about handwriting recognition, though this time recognizing handwriting scanned from sheets of paper, and using recurrent neural networks. |
| 1990-1 | Ingénieur de récherche at Laboratoire d'Informatique et Méchanique pour les Sciences de l'Ingénieur (LIMSI) in the Université de Paris Sud. There I worked on the European POLYGLOT project for continuous speech recognition using hidden Markov Models. |
| 1987-90 | MA in Mathematics and Electrical & Information Sciences, Trinity Hall, Cambridge University |