The Laureates

© Klaus Tschira Stiftung / Peter Badge

Alexei Efros

Born 9 April 1975, Sankt Petersburg, Russia

ACM Prize in Computing (2016) – For groundbreaking data-driven approaches to computer graphics and computer vision.

Alexei A. Efros was born in Leningrad, Soviet Union (today St. Petersburg, Russia), in 1975 and he spent his adolescence there. When he was 14, his father’s career path took the family to the United States. His family eventually settled down in Utah, and in 1997, Efros went on to receive his bachelor in computer science from the University of Utah.

It was in 1999, with a paper that Efros co-wrote with Thomas K. Leung, titled Texture synthesis by non-parametric sampling, that he created a buzz in the world of computer science. In texture synthesis, a small sample of a digital image is taken, an algorithm is applied to the sample’s structural content in order to enlarge the image, create background images, or perhaps fill in holes. Before their 1999 paper, synthesizing digital textures was done using complicated mathematics that produced aesthetically sub-par results for structured textures. In Efros’ words, “We tried to address this shortcoming by proposing a very different and extremely simple way of synthesizing textures locally, one-pixel-at-a-time. In recent years this algorithm has been used for synthesizing a large spectrum of textures as well as filling holes in textured regions.” The non-parametric modeling technique developed by Efros and Leung proved to be highly advantageous to the entertainment industry, especially in digital editing and 3-D graphics.

While working toward a PhD at UC Berkeley, Efros co-authored yet another significant paper with William T. Freeman published in 2001, Image Quilting for Texture Synthesis and Transfer. As explained by Efros, Image Quilting is when “a new image is synthesized by stitching together small patches of existing images.” A simple algorithm carries out texture transfer that creates a new object by borrowing the texture from another or “how an image can be re-rendered in the style of a different image.” He received his PhD two years later, in 2003.

In 2004, Efros made his way to Carnegie Mellon University and spent nine years on the faculty of the Robotics Institute. He would return to UC Berkeley in 2013 as an Associate Professor in the Computer Science Division in the Electrical Engineering and Computer Science Department (EECS). Efros is part of the Berkeley Artificial Intelligence Research Lab (BAIR), Computer Vision Group, and the Visual Computing Lab. He maintains affiliations with the University of Oxford and Team WILLOW at Laboratoire d’Informatique de l’École Normale Superiéure in Paris.
Efros’ prolificacy is not in question. He has written over 100 scholarly papers about various subjects, ranging from computer vision, computer graphics and artificial intelligence. Each significant breakthrough followed a seemingly logical progression in his work. The Scene Completion Using Millions of Photographs paper came along in 2008 in which Efros and James Hays presented an algorithm that patches an image’s holes with analogous photos selected from a database of millions of photos. Now, algorithms are regularly adapted to look through millions of social media photos for recognition research and processing.

Through his work with Richard Zhang and Phillip Isola, Colorful Image Colorization, Efros also developed an algorithm that automatically produces a color version of a black and white image. One million of images are analyzed to estimate the color of each part of the photograph as accurately as possible. In another paper with Phillip Isola, Jun-Yan Zhou and Tinghui Zhou, he was able to demonstrate that the exchange between two different networks can create unanticipated translations of an image; images drawn by one network are evaluated by the other to test how characteristic the image is.

The aforementioned projects, as well others where Efros has trained algorithms to analyze details and styles of millions of images, have opened worlds of possibilities in computer graphics and computer vision.

Efros received the 2016 ACM Prize in Computing, which was handed over at the annual ACM Awards Banquet on June 24, 2017 in San Francisco, California. Additional accolades and honors include: CVPR Best Paper Award (2006), NSF CAREER award (2006), Sloan Fellowship (2008), Guggenheim Fellowship (2008), Okawa Grant (2008), Finmeccanica Career Development Chair (2010), SIGGRAPH Significant New Researcher Award (2010), ECCV Best Paper Honorable Mention (2010), and the Helmholtz Test-of-Time Prize (2013).

Text adapted from: the Association for Computing Machinery (ACM) “Image Alchemist Alexei Efros to Receive ACM Prize in Computing” by Jim Ormond