Cordelia Schmid : Inria – French Académie des sciences Grand Prize

Cordelia Schmid joined the University of Karlsruhe in 1987 to begin a five-year course in Computer Science. During her Masters internship at the Grenoble Institute of Technology as part of an ERASMUS exchange, she decided to devote herself to computer aided vision, and more specifically, to object recognition. “I asked myself the following, very simple question: why is it so easy for humans to recognise objects, places, and people in an image, but almost impossible for a computer to do the same? I was interested in finding the scientific answer to this question. ”

A whole field of research to be explored

Cordelia Schmid first began developing a new form of image representation back when she started work on her PhD thesis, allowing objects to be recognised in realistic conditions (including concealment, rotation, and deformation). This early research already caught the attention of the scientific community. Later, she extended her research to visual learning. In combining research into computer aided vision and research into automatic image and video recognition learning, Cordelia Schmid became one of the most widely-acclaimed specialists in her field in the space of just a few years.

Since 2003, she has been leading a 20-strong project team at Inria: the Lear project, which changed its name to Thoth in 2016. “This next stage has enabled me to launch new projects and work more closely with other researchers. More people means faster progress. ”

Cordelia Schmid’s career has also had quite an international dimension, including a year as a post-doctoral student in Oxford and time spent in Berkeley, which she has visited regularly since 2000. “A researcher isn’t called to stay locked away in their office or lab. They can always benefit from interaction, taking about their own work, comparing it with that of other researchers, and receiving the latter’s constructive criticism. Going off to find out what’s happening elsewhere is essential. Especially if elsewhere happens to include the best research teams in the world. ”

Highly practical applications

When Cordelia Schmid began her research, computer aided vision was still in its infancy. There were no firm foundations; everything was as yet at an embryonic stage. At that time, computer aided visual recognition techniques were applied to very simple objects, taken from databases that ran to little more than a few hundred images. “We’ve come an incredibly long way. The field of machine visual recognition has grown exponentially. Today, research is focusing on far more complex content, using databases that can contain one hundred million images. What was beyond our wildest dreams is now within our grasp. ”

The list of potential applications for Cordelia Schmid’s research is set to become longer still; already, these range from autonomous vehicles to remote services for the elderly and children, as well as geolocation based on online images and videos. Security, CCTV, and health – with visual recognition applied to the realm of diseases – are yet other fields in which advances in research are having a direct impact. Internet giants such as Facebook and Google are also taking a keen interest in the rollout of these new, cutting-edge technologies.

Research: passion and perseverance required

Some twenty years on from her PhD, Cordelia Schmid acknowledges that research is still a valuable source of personal enrichment and fulfillment. “You’re constantly thinking about innovative projects, and always meeting people who are passionate about what they do. You never have time to get bored. Research is much more than a job; it’s a passion.” Schmid encourages young women and men working in research to “stay the course unwaveringly, despite the inevitable doubts, ” as she herself has: “I’ve always held onto the goals I’ve set myself, whatever the difficulties. Rewards don’t come overnight. They’re the result of a long journey – and recognition by the community is by no means journey’s end. ”

Testimonies

 

 

 

 

Jean Ponce, is a teacher at École Normale Supérieure and head of the Department of Computer Science at ENS, as well as the leader of the Inria-ENS-CNRS joint project team WILLOW.

"The field of computer aided visual recognition has grown rapidly since the early 2000s. Cordelia Schmid is one of the early pioneers behind this growth. Her work is outstanding. Already during her PhD thesis, she introduced an unvarying representation, opening up the way for considerable improvement in image classification. The algorithms she developed with her teams have won a great many international image recognition competitions. She has also devoted herself to establishing evaluation and comparison tools that have led to significant advances in the field. Cordelia has a strong scientific personality, as well as a clear view of the key issues: this is a very important quality in a researcher"

Stéphane Mallat, is a researcher in applied mathematics, and a teacher in the Department of Computer Science at École Normale Supérieure (Paris).
       

"I’ve known Cordelia Schmid since she defended her thesis in 1996. Her work has continued to have an impact on computer aided vision ever since. She successfully demonstrated that images could be identified and described in more complex settings, which was completely new at the time. She was one of the first people to conduct systematic campaigns for the stringent testing of visual recognition methods. These and many other similar contributions have led to Cordelia Schmid having a far-reaching influence, both through her research and through her ability to provide leadership in the scientific community. For instance, she is editor-in-chief of the International Journal of Computer Vision , one of the top two journals in the field."

Bio express

• 1992 - Master's degree in Computer Science at the University of Karlsruhe
• 1996 - PhD in Computer Science at the Grenoble Institute of Technology
• 1996-1997 - Post-doctoral student at Oxford University
• 2004 - Joined Inria as a researcher, later becoming research director
• 2006, 2014 et 2016 - Longuet-Higgins Prize
• 2015 - Humboldt Research Award