Benchmark instances for the min cost multicut problem

Abstract

This dataset contains problem instances of the min cost multicut problem (Chopra and Rao (1993)) which is also known as correlation clustering (Demaine et al. (2006)). The instances come from a variety of applications (image segmentation, modularity clustering, birdsong recognition). The origin of the instances is described in more detail in the readme.md file. Some of the instances are derived from graphical models of the opengm benchmark dataset (Kappes et al. (2015)). This benchmark dataset contain a wide variety of graph types (planar graph, sparse non-planar graphs, complete graphs).