Matthew Blaschko

Also known as Matthew B. Blaschko

Matthew Blaschko is a professor at KU Leuven who applies machine learning to visual data and directs ELLIS Unit Leuven, part of the European Laboratory for Learning and Intelligent Systems. He earned a summa cum laude doctorate from the Technische Universität Berlin in 2009 for work at the Max Planck Institute for Biological Cybernetics in Tübingen. His object-localization paper with Christoph Lampert and Thomas Hofmann, on efficient subwindow search, won the best paper award at CVPR 2008. After a Newton International Fellowship at Oxford, he joined the applied mathematics faculty of École Centrale Paris in 2011 and worked as a permanent research scientist at Inria Saclay. With Maxim Berman and Amal Rannen Triki, he presented the Lovász-Softmax loss for semantic image segmentation at CVPR 2018.

Now at: KU Leuven and European Laboratory for Learning and Intelligent Systems
Works: 320
Citations: 9,893
h-index: 38
Badges: Rising
Wayfarer

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

2005

Degrees at Columbia and UMass Amherst

He earns a B.S. from Columbia University and an M.S. from the University of Massachusetts Amherst. His 2005 master's thesis in the Department of Computer Science covers support vector classification of images with local features.
2008

Efficient subwindow search

With Christoph Lampert and Thomas Hofmann, he coauthors "Beyond Sliding Windows: Object Localization by Efficient Subwindow Search," which wins the best paper award at CVPR 2008. In 2008, his paper with Lampert on learning to localize objects with structured output regression wins the ECCV best student paper award, and their multiple kernel learning paper receives the DAGM 2008 main award.
2009

Doctorate on kernel methods

He completes a doctorate in electrical engineering and computer science, summa cum laude, at the Technische Universität Berlin in 2009 for work done at the Max Planck Institute for Biological Cybernetics in Tübingen. His dissertation, "Kernel Methods in Computer Vision: Object Localization, Clustering, and Taxonomy Discovery," has Bernhard Schölkopf, Klaus-Robert Müller and Thomas Hofmann as advisors.
2009–2011

Newton fellow at Oxford

As a Newton International Fellow in the University of Oxford's Department of Engineering Science, he works in the Visual Geometry Group, where a reading group on structured output learning meets in 2009. With Andrea Vedaldi and Andrew Zisserman, he presents work on simultaneous object detection and ranking with weak supervision at NIPS 2010.
2011–2014

Faculty at École Centrale Paris

He joins the applied mathematics faculty of École Centrale Paris in October 2011 and works as a permanent research scientist at the Inria Saclay research center. His students in France include Wojciech Zaremba, who completes an École Polytechnique master's thesis in 2012, and Katerina Gkirtzou, who earns an École Centrale Paris doctorate in 2013. In 2014, he receives a habilitation from the École Normale Supérieure de Cachan.
By 2017

Professor at KU Leuven

He moves to KU Leuven, where he becomes a professor in the Processing Speech and Images group of the Department of Electrical Engineering and applies machine learning to visual data. His doctoral student Wacha Bounliphone completes a joint Université Paris-Saclay and KU Leuven doctorate in 2017 on statistically and computationally efficient hypothesis tests for dependency and similarity.
2018

The Lovász-Softmax loss

With Maxim Berman and Amal Rannen Triki, he presents the Lovász-Softmax loss at CVPR 2018, a method for directly optimizing the mean intersection-over-union loss of neural networks in semantic image segmentation. The loss raises segmentation scores on the Pascal VOC and Cityscapes datasets.
As of October 2026

Director of ELLIS Unit Leuven

He is a professor at KU Leuven and heads the Processing of Speech and Images subdivision of its electrical engineering department. He directs ELLIS Unit Leuven, established in January 2019 within the European Laboratory for Learning and Intelligent Systems, and is a fellow in the ELLIS Health program. He is also a core PI in the Flanders AI Research Program, leading work packages on decision support systems and medical imaging.