Diverse M-Best Solutions by Dynamic Programming

Abstract

Abstract. Many computer vision pipelines involve dynamic programming primitives such as finding a shortest path or the minimum energy solution in a tree-shaped probabilistic graphical model. In such cases, extracting not merely the best, but the set of M -best solutions is useful to generate a rich collection of candidate proposals that can be used in downstream processing. In this work, we show how M -best solutions of tree-shaped graphical models can be obtained by dynamic programming on a special graph with M layers. The proposed multi-layer concept is optimal for searching M -best solutions, and so flexible that it can also approximate M -best diverse solutions. We illustrate the usefulness with applications to object detection, panorama stitching and centerline extraction.

Efficient Vertex-Oriented Polytopic Projection for Web-Scale Applications

“…Because ranking of items needs to be obeyed and there are edge costs between consecutive items in a slate, dynamic programming can be used to give an efficient polynomial time Algorithm (Keerthi and Tomlin 2007) for Step 1. For incrementally finding of the next best slate (Step 3), the ideas in Haubold et al (2018) can be used to give efficient polynomial time algorithms. It is clear from these specializations that Algorithm 2 is powerful and can be potentially applied more generally to problems with other complex structures.…”