Combinatorial Image Analysis
Our research group solves open problems in image analysis by means of combinatorial optimization techniques.
Toward this goal, we devise abstractions of image analysis tasks in the form of combinatorial optimization problems, we study their properties, and we adapt and design algorithms to solve these problems. In this effort, we strive for exact (globally optimal) solutions. Where exact solutions are out of reach, we aim at approximations with guaranteed bounds and certificates.
Research Focus: Minimum Cost Multicuts in Image Analysis
Minimum Cost Multicuts for Image Segmentation
Keuper et al. 2015
Andres et al. 2012
Andres et al. 2011
Minimum Cost Multicuts for Multi-Target Tracking
Jug, Levinkov et al. 2015
Tang et al. 2015
Minimum Cost Multicuts for Human Body Pose Estimation
Pishchulin et al. 2016
Lifting of Multicuts