Think Like an Amateur, Do As an Expert: Lessons from a Career in Computer Vision

Tuesday, May 22, 9:00 AM - 10:30 AM
Mission City Ballroom B2-B5

Dr. Takeo Kanade will share his experiences and lessons learned in developing a vast range of pioneering computer vision systems and autonomous robots, including face recognition, autonomously-driven cars, computer-assisted surgical robots, robot helicopters, biological live cell tracking and a system for sports broadcasts.

Most researchers, when asked their fondest desire, respond that they want to do good research. If asked what constitutes “good research,” they often find it difficult to give a clear answer. For Dr. Kanade, good research derives from solving real-world problems, delivering useful results to society. “Think like an amateur, do as an expert” is his research motto: When conceptualizing a problem and its possible solution, think simply and openly, as a novice in that field, without preconceived notions. When implementing a solution, on the other hand, do so thoroughly, meticulously and with expert skill.

In his research projects, Dr. Kanade has met and worked with people from diverse backgrounds, and has encountered many challenges. While exploring the technical side of some of his most important projects, he will also describe experiences that highlight the enjoyable aspects of a researcher’s life—those that have occurred accidentally or inevitably as his “Think like an amateur, do as an expert” approach has guided his interactions with problems and people.


Dr. Takeo Kanade

U.A. and Helen Whitaker Professor, Carnegie Mellon University

From his early work on face recognition, to the 3D imaging techniques he developed for the 2001 Super Bowl broadcast, to his work on computer-assisted surgical systems for hip replacement, Dr. Takeo Kanade is one of the world’s foremost researchers in computer vision and robotics. He has made important contributions in a number of areas, ranging from foundational theoretical advances to innovative devices and algorithms—including techniques for facial expression analysis that can spot minute details and changes.

Kanade pioneered key techniques in motion analysis, introducing a fundamental algorithm for tracking image patches through a video sequence, opening the door for modern tracking and compression technologies. He also developed the concept of “virtualized reality” in which several images of an event taken from various vantage points are combined into a coherent model which can be then be used to create novel views of the scene from any desired viewpoint.

Kanade’s innovations in robotic vision have been incorporated into several autonomous robots, including human-face recognition, a driverless car, biological live cell tracking through a microscope and an autonomous helicopter.

Along with many other honors, Dr. Kanade received the 2016 Kyoto Prize, Japan's highest private award for global achievement.

[Portions courtesy of the Inamori Foundation.]

Join us May 21-24, 2018 in Santa Clara, California.