# From Coarse to Fine: Robust Hierarchical Localization at Large Scale

## TL;DR

This paper introduces HF-Net, a hierarchical localization system that combines global retrieval and local feature matching using a CNN, enabling real-time, robust large-scale 6-DoF localization despite appearance changes.

## Contribution

The paper presents HF-Net, a novel CNN-based hierarchical localization method that improves robustness and efficiency for large-scale environments, outperforming existing approaches.

## Key findings

- Achieves state-of-the-art accuracy on large-scale localization benchmarks.
- Enables real-time operation with significant runtime savings.
- Demonstrates robustness to large appearance variations.

## Abstract

Robust and accurate visual localization is a fundamental capability for numerous applications, such as autonomous driving, mobile robotics, or augmented reality. It remains, however, a challenging task, particularly for large-scale environments and in presence of significant appearance changes. State-of-the-art methods not only struggle with such scenarios, but are often too resource intensive for certain real-time applications. In this paper we propose HF-Net, a hierarchical localization approach based on a monolithic CNN that simultaneously predicts local features and global descriptors for accurate 6-DoF localization. We exploit the coarse-to-fine localization paradigm: we first perform a global retrieval to obtain location hypotheses and only later match local features within those candidate places. This hierarchical approach incurs significant runtime savings and makes our system suitable for real-time operation. By leveraging learned descriptors, our method achieves remarkable localization robustness across large variations of appearance and sets a new state-of-the-art on two challenging benchmarks for large-scale localization.