Recent efforts in the area of joint object matching approach the problem by taking as input a set of pairwise maps, which are then jointly optimized across the whole collection so that certain accuracy and consistency criteria are satisfied. One natural requirement is cycle-consistency – namely the fact that map composition should give the same result regardless of the path taken in the shape collection. In this talk, we introduce a novel approach to obtain consistent matches without requiring initial pairwise solutions to be given as input. We do so by optimizing a joint measure of metric distortion directly over the space of cycle-consistent maps; in order to allow for partially similar and extra-class shapes, we formulate the problem as a series of quadratic programs with sparsity-inducing constraints, making our technique a natural candidate for analyzing collections with a large presence of outliers. The particular form of the problem allows us to leverage results and tools from the field of evolutionary game theory. This enables a highly efficient optimization procedure which assures accurate and provably consistent solutions in a matter of minutes in collections with hundreds of shapes.

Dr. Luca Cosmo received his PhD in Computer Science from Ca' Foscari University of Venice in February 2016 where he currently holds a PostDoc position. His research interests spread among different aspects of Computer Vision and Pattern Recognition. During his Phd he mainly focused on the problem of camera calibration and three-dimensional acquisition. In 2015 he spent a period at TU Munich under the supervision of Prof. Daniel Cremers, where he started working on the problem of Shape Analysis. Since 2014 he is collaborating with a Computer Vision spin-off, DigitalViews S.r.l, in some industrial projects concerning the in-line quality control of silkscreen printings and conditioned air pipes. He is co-founder of DigitalMetrix S.r.l, a company operating in the field of three-dimensional acquisition for quality control purposes.