Alex Flint is a DPhil candidate with the Active Vision Lab at Oxford University, where his work focusses on developing algorithms to understand the world as seen through a camera. In 2008 Alex was awarded a scholarship from the Clarendon Fund in order to undertake his DPhil. This scholarship is awarded for overall academic merit and accepts less than 7% of applicants. Alex completed his undergraduate degree in Computer Science at the University of Adelaide, from which he has subsequently received a University Medal for outstanding academic achievement. After graduating in 2007 he worked for Google for six months before beginning his DPhil at Oxford. In 2010 he founded the Oxford Transhumanist society to promote discussion and debate on issues around rationality, ethics, and technology. He also has been a member of the student committee for the Oxford Entrepreneurs and has helped found the Clarendon Scholars Council.
Ogmento
I have accepted a computer vision engineering role at Ogmento beginning April 23rd.
MIT
I will be away from Oxford until mid-September as I am a visiting student with Bill Freeman at MIT in Cambridge, Massachusetts.
PhD Research
I am a PhD student with the Active Vision Lab and a member of Exeter College at Oxford University in England. Recent publications are available here.
The elevator pitch. My research concerns computer algorithms that understand the visual world as seen through a camera in sufficient detail to provide useful real-time information to humans.
The whole story. Since its inceptions the field of multiple view geometry within computer vision has mostly concerned itself with fast, accurate, and detailed 3D reconstructions. My research aims to leverage these geometric ideas instead for scene understanding, which has until recently been the exclusive domain of monocular (single-image) computer vision. To this end I work with indoor Manhattan environments in which the world is built up from axis-aligned surface. I have developed a fully Bayesian model for inference in this domain, which happens to have a very fast (100ms) dynamic programming solution. For more details see the publications section.
The elevator pitch. My research concerns computer algorithms that understand the visual world as seen through a camera in sufficient detail to provide useful real-time information to humans.
The whole story. Since its inceptions the field of multiple view geometry within computer vision has mostly concerned itself with fast, accurate, and detailed 3D reconstructions. My research aims to leverage these geometric ideas instead for scene understanding, which has until recently been the exclusive domain of monocular (single-image) computer vision. To this end I work with indoor Manhattan environments in which the world is built up from axis-aligned surface. I have developed a fully Bayesian model for inference in this domain, which happens to have a very fast (100ms) dynamic programming solution. For more details see the publications section.
Rationality
In 2010 I founded a student society called the Oxford Transhumanists to promote discussion on rationality, ethics, technology, and the future. We have since hosted many prominent speakers and we now hold friendly weekend meetups each Sunday at 6pm. To find out more or get involved, join the rationality Google Group. You may also be interested in:
Oxford Ideas
In 2010 four Oxford PhD students including myself founded Oxford Ideas. Our first project is an application for smart phones that provides relaxing stories to help people calm their mind before an important meeting, presentation, or exam. We ran a university-wide story-writing competition, which was judged by author Phillip Pullman, author of His Dark Materials trilogy. Stay in touch on the Oxford Ideas facebook page.
Blog
My blog is called "Living Dymaxion" as a tribute to innovation, productivity, and thinking-outside-the-box engendered by Buckminster Fuller. Unfortunately, it hasn't received enough attention in the last few months. I also contribute to the community rationality project LessWrong.
