A Simple Hierarchical Pooling Data Structure for Loop Closure

TL;DR

This paper introduces a simple hierarchical pooling data structure that significantly accelerates large-scale loop closure tasks by averaging descriptors, matching the performance of more complex methods at a fraction of the computational cost.

Contribution

The paper presents a novel hierarchical averaging data structure for loop closure that is simple, efficient, and achieves speedups comparable to sophisticated methods.

Findings

Achieves 4 to 20 times speedup on benchmark datasets.

Maintains comparable accuracy to complex agglomerative schemes.

Minimal performance loss despite simplicity.

Abstract

We propose a data structure obtained by hierarchically averaging bag-of-word descriptors during a sequence of views that achieves average speedups in large-scale loop closure applications ranging from 4 to 20 times on benchmark datasets. Although simple, the method works as well as sophisticated agglomerative schemes at a fraction of the cost with minimal loss of performance.