Cordelia Schmid is an Inria researcher in the field of computer vision, and more specifically in visual recognition associating invariant image descriptors and learning methods. Her research allows a computer to learn to interpret all types of real images and videos, to recognize objects, but also actions and places, and to index large image and video databases (more than one hundred million images). It is thus among the precursors and world leaders of modern visual recognition methods. In 2012, Cordelia Schmid received the prestigious IEEE Fellow award and an ERC Advanced Grant.
"The next challenge is the indexing of ever-larger image and video databases by poorly supervised methods, i.e. with less and less human intervention. This is the subject of the ERC Advanced Grant ALLEGRO (active large-scale learning for visual recognition), which aims at the independent learning of visual concepts from the enormous amount of data available on the Internet. The idea is to use the additional data associated with these images and videos, such as written annotations, oral comments, scripts, to save the tedious and potentially incomplete work of current methods of supervised learning. The challenge will be to be robust to the heterogeneity, relative inconsistency and highly variable quality of the information available," says Cordelia Schmid.
Key dates
- 1996 : PhD at the National Polytechnic Institute of Grenoble entitled "Local Greyvalue Invariants for Image Matching and Retrieval" (Prize for the best thesis of the INPG)
- 1996 – 1997 : postdoctorate at Oxford (Robotics Research Group )
- 1997 : researcher at the Inria Grenoble Rhône-Alpes centre
- 2003 : creation and responsibility of the LEAR project team
- 2004 : INRIA senior researcher
- 2012 : IEEE Fellow and ERC grant
- 2013 : ERC advanced Grant ("Active Large-scale LEarninG for visual RecOgnition ")
- 2014 - 2015 : Thomson Reuters Highly Cited Researcher.
