Efficient Pose and Cell Segmentation using Column Generation

TL;DR

This paper introduces a novel column generation approach for efficient multi-person pose and cell segmentation in images, formulating these tasks as integer programs and solving them via small-scale exact optimizations.

Contribution

It presents a generic relaxation scheme using column generation for combinatorial pose and cell segmentation problems, improving efficiency.

Findings

Efficient exploration of pose and cell spaces.

Effective integer program relaxation scheme.

Applicable to natural and biological images.

Abstract

We study the problems of multi-person pose segmentation in natural images and instance segmentation in biological images with crowded cells. We formulate these distinct tasks as integer programs where variables correspond to poses/cells. To optimize, we propose a generic relaxation scheme for solving these combinatorial problems using a column generation formulation where the program for generating a column is solved via exact optimization of very small scale integer programs. This results in efficient exploration of the spaces of poses and cells.