Visual Place Recognition (VPR) is the ability to correctly recall a previously visited place using visual information under environmental, viewpoint and appearance changes. An emerging trend in VPR is the use of sequence-based filtering methods on top of single-frame-based place matching techniques for route-based navigation. The combination leads to varying levels of potential place matching performance boosts at increased computational costs. This raises a number of interesting research questions: How does performance boost (due to sequential filtering) vary along the entire spectrum of single-frame-based matching methods? How does sequence matching length affect the performance curve? Which specific combinations provide a good trade-off between performance and computation? However, there is lack of previous work looking at these important questions and most of the sequence-based filtering work to date has been used without a systematic approach. To bridge this research gap, this paper conducts an in-depth investigation of the relationship between the performance of single-frame-based place matching techniques and the use of sequence-based filtering on top of those methods. It analyzes individual trade-offs, properties and limitations for different combinations of single-frame-based and sequential techniques. The experiments conducted in this study demonstrate the benefits of sequence-based filtering over the single-frame-based approach using various VPR techniques. We found that applying sequence-based filtering to a lightweight descriptor can enable higher VPR accuracy than state-of-the-art methods such as NetVLAD, while running in shorter time. For example, matching a sequence of 16 images, CALC descriptor outperforms NetVLAD on Campus Loop dataset while taking about 22% less time to perform VPR.@en

Visual place recognition (VPR) is the process of recognising a previously visited place using visual information, often under varying appearance conditions and viewpoint changes and with computational constraints. VPR is related to the concepts of localisation, loop closure, image retrieval and is a critical component of many autonomous navigation systems ranging from autonomous vehicles to drones and computer vision systems. While the concept of place recognition has been around for many years, VPR research has grown rapidly as a field over the past decade due to improving camera hardware and its potential for deep learning-based techniques, and has become a widely studied topic in both the computer vision and robotics communities. This growth however has led to fragmentation and a lack of standardisation in the field, especially concerning performance evaluation. Moreover, the notion of viewpoint and illumination invariance of VPR techniques has largely been assessed qualitatively and hence ambiguously in the past. In this paper, we address these gaps through a new comprehensive open-source framework for assessing the performance of VPR techniques, dubbed “VPR-Bench”. VPR-Bench (Open-sourced at: https://github.com/MubarizZaffar/VPR-Bench) introduces two much-needed capabilities for VPR researchers: firstly, it contains a benchmark of 12 fully-integrated datasets and 10 VPR techniques, and secondly, it integrates a comprehensive variation-quantified dataset for quantifying viewpoint and illumination invariance. We apply and analyse popular evaluation metrics for VPR from both the computer vision and robotics communities, and discuss how these different metrics complement and/or replace each other, depending upon the underlying applications and system requirements. Our analysis reveals that no universal SOTA VPR technique exists, since: (a) state-of-the-art (SOTA) performance is achieved by 8 out of the 10 techniques on at least one dataset, (b) SOTA technique in one community does not necessarily yield SOTA performance in the other given the differences in datasets and metrics. Furthermore, we identify key open challenges since: (c) all 10 techniques suffer greatly in perceptually-aliased and less-structured environments, (d) all techniques suffer from viewpoint variance where lateral change has less effect than 3D change, and (e) directional illumination change has more adverse effects on matching confidence than uniform illumination change. We also present detailed meta-analyses regarding the roles of varying ground-truths, platforms, application requirements and technique parameters. Finally, VPR-Bench provides a unified implementation to deploy these VPR techniques, metrics and datasets, and is extensible through templates.@en