Convolutional Neural Network-based Place Recognition

TL;DR

This paper introduces a CNN-based place recognition method that combines learned features with spatial and sequential filtering, achieving significant improvements on benchmark datasets over previous techniques.

Contribution

It is the first to apply CNN features to place recognition, integrating spatial and sequential filters, and provides a comprehensive analysis of features from all CNN layers.

Findings

75% increase in recall at 100% precision on benchmark dataset

Outperforms previous state-of-the-art methods significantly

Analyzes features from all CNN layers for robustness

Abstract

Recently Convolutional Neural Networks (CNNs) have been shown to achieve state-of-the-art performance on various classification tasks. In this paper, we present for the first time a place recognition technique based on CNN models, by combining the powerful features learnt by CNNs with a spatial and sequential filter. Applying the system to a 70 km benchmark place recognition dataset we achieve a 75% increase in recall at 100% precision, significantly outperforming all previous state of the art techniques. We also conduct a comprehensive performance comparison of the utility of features from all 21 layers for place recognition, both for the benchmark dataset and for a second dataset with more significant viewpoint changes.