Researchers Push Towards New Frontiers in Computer Vision

A poster announcing the Frontiers in Computer Vision Workshop. October 11, 2011 In August, researchers from academia and industry gathered to discuss the past, present and future of the field of computer vision during the Frontiers in Computer Vision Workshop at the Stata Center. Unlike many academic workshops, where the overriding goal is to facilitate discussion and collaboration on innovative research, the Frontiers in Computer Vision Workshop was designed to help define a clear goal for the field that will lay the groundwork for the next generation of computer vision research.   CSAIL Principal Investigator Aude Oliva, who co-organized the workshop with Professor Alan Yuille of UCLA, explained that the workshop more than exceeded her expectations. Of the 90 people invited to attend, all eagerly gathered for three days of sessions covering everything from the field’s history to key research advances and industry applications. Most importantly, explained Oliva, researchers decided that a key mission going forward is providing better education for the next generation of computer vision researchers, which they hope will help propel the field forward.   “We really packed 14 different sessions in three days. The sessions covered scientific issues and issues related to not only where the field should go, but also what is the historical perspective of the field, what are the challenges and the questions that are important to look at,” said Oliva.     Since the last workshop on the challenges and directions in computer vision 20 years ago, researchers have seen the rise of the Internet, an innovation that has made a major impact on research developments in the field. Computer vision started with the goal of building machines that can see like humans, but it has expanded to include applications such as online image searching, computational photography, biological imaging, vision for nanotechnology and much more. By providing access to vast quantities of information, the Internet has allowed computer vision algorithms and methods to progress dramatically.   The Frontiers in Computer Vision Workshop this past summer provided participants a glimpse of research activities underway in the field, as well as a chance to connect with colleagues, new and old alike. From feedback Oliva and Yuille have received, it seems the workshop has already to begun to spark seeds of interest in new areas of research and collaboration.   According to Oliva, many participants relayed that, “it was great to have a view, the gist, the summary of what everybody does in three days.”   “We discussed the taxonomy, the scholarship, the relations between biological vision and computer vision, but what was really important was to see the different areas where people are excited and see some of the challenges and the questions going forward,” she added.   An important outcome of the workshop was a joint decision by participants to focus their efforts on training and educating the next generation of computer vision researchers. Participants expressed the need for creating summer school opportunities for students interested in exploring the field, developing comprehensive textbooks, and increasing awareness of the field, for instance through Wikipedia, in addition to posting educational videos and tutorials online. Since the workshop, participants have continued the discussion and have commenced compiling a list of 20 different techniques in computer vision that they feel every researcher should learn.   “Something that was discussed a lot is how to train the next generation and what students need to know. People came up with a list of the top questions and most important methods that we should teach to students, to the next generation of computer vision researchers and engineers,” said Oliva. “I think this was one of the most important outcomes of the workshop, all of us agreeing that we need to come up with training and education strategies because then the field can grow and cross-fertilize if people know all these different techniques and methods.”   Since the workshop, participants have been actively engaging and collaborating on ideas to further advance the field. Oliva and Yuille are compiling a report on the workshop, which will guide the next steps, including organizing a second frontier workshop at the IEEE’s Computer Vision and Pattern Recognition Conference in 2012.   “We hope this will have a snowball effect,” said Oliva. “The idea is that we started it as a little flame and now other people have to spread it.”   Looking back on the workshop, Oliva hopes that its legacy will be to lay the groundwork for a better future for the field.   “If 20 years from now the younger generation says, ‘Here is what the Frontiers in Computer Vision Workshop in 2011 did, they changed the field because they decided people needed to be well trained and educated in the various inter-disciplinary methods,’ then that will be the most rewarding outcome,” said Oliva.   For more information on the Frontiers in Computer Vision Workshop, which was funded by the National Science Foundation and the U.S. Army Research Office, please visit: http://www.frontiersincomputervision.com/. Abby Abazorius, CSAIL