I am a PhD candidate at the School of Computer Science, University of Waterloo, Canada. My supervisor is Yuri Boykov.
My research is focused on designing general unsupervised and semi-supervised methods for accurate image segmentation and object delineation. In particular, I am focused on segmentation/detection of thin objects of different nature, e.g. roads, blood vessels, or contrast edges. The problem is formulated as cost function minimization. The concrete choice of the function depends on the properties of objects of interest. Many useful formulations are NP-hard, therefore efficient approximate optimization algorithms for such functions are desired.
https://github.com/dmitrii-marin