Engineer Takeo Kanade (Tamba, Hyogo, Japan, 1945) is the 2024 BBVA Foundation Frontiers of Knowledge Award winner in its Information and Communication Technologies category for his development of the mathematical foundations on which the capabilities that allow computers and robots can "understand and interpret visual images", according to the jury. The awarding of the prize takes into consideration that a multitude of practical applications of artificial vision, such as robotic surgery, facial identification of mobile phones, autonomous vehicles and replays of plays from multiple angles in sports broadcasts, are possible today due to the contributions de Kanade, who has been a professor of Computer Science and Robotics at Carnegie Mellon University (Pittsburgh, United States) since 1980.
The jury considered that Takeo Kanade's contributions "have not only shaped the scientific disciplines of artificial intelligence and robotics, but have also significantly transformed the technological world in which we live." The winner, who has declared that he feels "very honored to have been selected for the prestigious Frontiers of Knowledge Award" and to add his name to the list of illustrious winners in previous editions. Shortly after being informed of the award of the award, the engineer has indicated that, “as demonstrated by the fact that the visual cortex occupies the dominant area of the human brain, vision or the processing of visual information provides humans with the richest and most important information channel for understanding and communication. ". "Artificial intelligence and robots with similar or even better computer vision capabilities contribute to improving our lives. I see many opportunities," he noted.
Kanade's contributions revolutionized three-dimensional computer vision by developing algorithms much faster than those that existed at the time. To have a three-dimensional image, two cameras are needed, which processed the information separately, but Kanade changed the way of doing it. It was much faster to take advantage of the information about the movement of objects recorded by each camera to understand how the image moves even before integrating the videos from all the cameras. “Once we understand this, we no longer need to send all the color or video information, but it is enough to simply send the movement,” explained the engineer.
Kanade developed together with his doctoral student Bruce Lucas a new method to estimate optical flow, the method known as Lucas-Kanade, which captures the shapes of objects and deduces their speed and direction of movement. “That is the basis of video coding, and my optical flow algorithm is used for practically any moving image data compression technique,” indicated the professor, who also developed a way to simplify the calculations made by the computer to process the images. This advance, together with doctoral student Carlo Tomasi, was published in 1992 in the International Journal of Computer Vision and made it possible for computers of that time to already be able to work with three-dimensional images.
Thanks to these advances, two researchers from Carnegie Mellon University, the same one where Kanade is a professor, traveled across the United States from coast to coast on highways in one of the first autonomous vehicles without barely touching the steering wheel, only the brake and accelerator. . Another practical application in the audiovisual world came with the 2001 Super Bowl final, where the EyeVision technique was used for the first time, with several cameras that allowed any scene to be reproduced in 360 degrees in motion and from any point of view. none of the cameras were there. In Spain it has been used to watch replays of goals from different angles, even from the point of view of the scorer or the ball itself. The technique allows the Hawkeye feature in tennis to know exactly where the ball hits the ground.
